and... .1- 5.. 514.. Q.
.. .. “Jan. "1... .u» "out...

S.WW..:P . Jfinrfimnwwfi

.. W1...$..40.4.§.&W Kfigriw

”elk .. . ( . HECTVR‘ I.

.12. I t. in}... ban L- 5.1.:

S M?“ 5.5...

 

   

u
x.
V V,-
v‘

n ..I\.

   

flonvshtu 3L at...
«A :3 .0 1.. 0
.... . . 61".. .
. . , .(

 

 

     

 

 

: .
. . .
N . C l"'.' v.
I ‘ Q... ‘I._‘ C.IC~QD> Q .0.
. 29L... . .
- . .3}.d.k.d
.I 4.310%”..4- .
1 o. nfl‘c o q . .
‘ C A '
a 9...»... N: .3. v
.2. ...3}...v..1...
: . .. . .
Au .s.

 

    

                      
             

13%;: I
'13 n .
.11. ~-

I
k
' .1...

'p i‘mm

.1 .
1mm

. ';= 'islw‘

{ 4'

O.II
0 ..0O’IV1Q o

‘ ‘
v

5 t‘ ‘ .
utotw-A , ‘

' 31 vac-.14 ‘ . t“ -30‘.‘
"v. ..¢-",A _u A
V..'§v“‘1 n". v'
I ‘.lb ‘1'., t‘l

. .; . ; ,v
‘v‘..l Iy‘ l‘". 0
' ‘5 l
n
’ 'I‘

                      

% y . .-
.lu. .. ,
II .t .blflfll‘. |I-t).|
tn -1. il“l\ll‘l‘|\l'oo:!li
'0! 1|ll||lrI|$ ~|LIFI|II 1.)
It]. .altlctll'l‘lu‘u-v.lllilv|n\ll ia&§.hrloéyz.il. «‘2
n - 1 an
.1ID\.I»IJI' I‘lln‘l .llsudl‘zwul‘l‘o 3&1: illicit
I" '.1‘1‘1I‘0.%01‘01~4$
‘0 . 'll.’ - -
\1...It. .. r‘toll... I.\.|< .prw'h. II...I\IL.'ou|.I\Jn\VI .1!!! Ext-law“... . 1
‘. Out. ‘1 - O - - v| D ‘Iall I
.llllxl.‘ .\l$111{ltlit!r»t\£ 1‘}... \l‘nlicl‘iul‘. .I\Il.I:o:v0§l . . . . . J! , J..\ .2 in... v
I V r. u--‘l| ! n .I .v— u u in. o vl' I l .I Jnu‘.‘ VI.” ’. IQIIIII’ N l
n o I. ~\1>\ \v-u H.¢Ir. t... I u l 14!". 3. 4| IvH|VIIEWI|Cllr
14.1‘vv |.I« {.0 .0) . v-II C» u >3 VI.I t . It? AI”...U.HI [Ill-“h
Ia .ln ‘ at I x. v . .~‘AIA 1|" 0
v‘ .I‘ 0.0! ‘v 5 IO 0 n | I 11‘444 1”“ 0 l1 ‘1
. .a .24.... . 32... . . .1 ‘ r..- . . . .- . lxh...!d.~1flfll.l.ilu€.lull\ulll1l|l
. . .1. .0. o t v . .. . u- lv.|-1.l- I qt 11".
-Q I n '.b ‘ v v . «V A: n - -‘Il!’09\l\ \ll‘ ll.’ 0"
9 L I u I tn 5 -.1 It ‘. Ilixi“ '.‘bll.'|
I. - '1.‘tl1.lu.|tlz.110 thin-
- .‘I‘u V‘V 1“-.|4ID§‘O\'I
. u... o‘ll.|§ll.lu‘ .I.AI
.. nl‘Jll'J . clvclb. uo. nofl
- l1.l.|t-. .. ..¢.| 1: ||cr -
4.-l\ .-. :11.le
. \‘Iltlol 2’1! -1..-
. If. '9' R 11 . . l
. y to {JR-FM“ Ylofluunntlfl.‘ It‘lul
. ‘9‘.-.s.' -un'bkfllul
. . .1 «By 1.- . lxll.(LI-.| I
- . qlul II‘II..I lvl|ocll§lu 1:
. rl‘d.\ ‘II’IIIIO n1lw» ll
‘0‘? tia‘v-0‘1uv.|1
\ .lllscl t! n 3:.» u‘ll.‘NII {I'lvut
..I 11-1.?1...‘ .ILIHI. :o.'\.
1". .I ..\.v.|l . «I. . 3..-
.- . .\....A.-:IH.. In} .LV
1 ...I1. .- §‘ .‘II
c 'll‘v.'.l..|9tr...c .I-
0 l‘»....‘..1!u I'll‘ ..
1.. .II|:I.1\-$§]O\.OI\‘- \.
- v .. . . v. -IXvI. .111u?‘ -
a I1|3¢K.‘III
. ..v .. a x 1 . . hill}. .ln..h
31301:. . in. . 5... I‘ll- '3'
. .nb-o.\..\l|.lt.|‘.t1 b. ‘I
-..r\‘|).|nl {I‘k llu‘ ul‘ilPiJ
ul‘f‘41ctn- KID 9!. . .I\ v.0 {‘3 ll.
4.. . .. L' 41‘nln1lliil‘nn‘l'un
.. in 1. ”PH‘
. \I

L. .10.»;
Z.IIV: on...
n

     

‘l'- t
4 I‘-

. . u < .- .-
. .
.J -- .p... o
. u \‘tl. 4101‘.H1Hh.h...£..l.\ . . v
‘ v. .. ‘Il'l. I Inn“?
.0. ~IA I 1 I at“ n
.
.nn

       

 

\Iul!!!
‘12..
4315.} lllLHWw.‘ - .. -. .
4 J‘I..v4 v .‘ . l A I
| o (It til-u Oltni.n.|'. Av... . I :0
(Lilinlrvrlll: fifl- a-. ..
1.1.. 1?]... I». , .. .
. II -tivill . , .flfiu -. -
. . -u.l»\la.(.. Huh») .2 (:9...lt.|~..9..!..t- !.
o‘sul..-‘ § "Du‘D 1‘
v.11... . gang-t. C 3- 411.41.31.11.
:h .- g. . - .uunlli t
--.r.ll 3“: 3‘. «mpg .uUklRt‘klfcllltk I‘llllcll. . -
» . . u-. .{ . tin .- - . - no-.- uni. .<1I("I»..ll. !J\-- \ - I l‘l....|

  

Au \ II‘!
n .
no.1 . ¢

W”! vfi“: .‘Vlfll. .|.

    

'1. a

 

llllHlHllHJllHllHillll”IIIIUHIHHIWHIIJ‘. Will.

THESIS 93 10423 5159 .- L/I

 
  

LIBRAR Y
Michigan State

This is to certify that the

thesis entitled

A SPATIAL ANALYSIS OF LAKE SUPERIOR SHIPWRECKS:
A STUDY IN THE FORMATIVE PROCESS OF THE ARCHAEOLOGICAL RECORD

presented by

CHARLES ALLEN HULSE

has been accepted towards fulfillment
of the requirements for

Ph. D. degree in Anthrogology

AWL

Major professor

 

 

 

Date August 3, 1981

 

0-7 639

‘ OVERDUE FINES:
N- 25¢ per day per item
. l '
‘ j [fl-‘Mfi } , RETURNING LIBRARY MATERIALS:
,\ .113“ a” ' ' Place in book return to nemove
“MI" . charge from circulation records

 

 

 

 

A SPATIAL ANALYSIS OF LAKE SUPERIOR SHIPWRECKS:
A STUDY IN THE FORMATIVE PROCESS OF THE
ARCHAEOLOGICAL RECORD

By
Charles Allen Hulse

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Anthropology

1981

Thi
formative p
Lakes shipw
shipwrecked
these three
attribute r
comprise th
Such variab
frequency 0
SPECific ge
Spatial pat

The
imPortant x,
formation 5

comodifies

 

random pat:
Patterning
that there
indusuy a.

VIQCks. I"

then discr.

 

"I‘ll, 1‘"
I w)

24

.'I p
I ‘.‘
sun
o'I/l M.
' u

ABSTRACT
A SPATIAL ANALYSIS OF LAKE SUPERIOR SHIPWRECKS:

A STUDY IN THE FORMATIVE PROCESS OF THE
ARCHAEOLOGICAL RECORD

By

Charles Allen Hulse

This study describes a research effort which focuses upon the
formative process of the archaeological record as it applies to Great
Lakes shipwrecks, this research analyzes the spatial distribution of
shipwrecked vessels from.the iron, grain, and coal industries. Using
these three commodities a model of hypothesized distributions and
attribute relationships is created. A series of twenty-one hypotheses
comprise this model and are tested within the body of this research.
Such variables as vessel type, date of loss, commodity, weather,
frequency of traffic, etc. are tested for their association.with
specific geographical locations in order to evaluate their effects upon
spatial patterning of Lake Superior shipwrecks.

The testing phase of this research revealed a number of
important variable associations that have a significant effect upon the
formation of the archaeological record. For example, of the three
commodities under consideration only the iron industry exhibited none
random patterning of related shipwrecks. The interpretation of this
patterning in combination with results of other hypotheses tested is
that there is a direct correlation between the capital intensity of an
industry and the respective spatial distributions of associated ship-
wrecks. The causitive mechanisms which result in this correlation are

then discussed as they apply to the formation of the shipwreck

archaeol
account
and the
transt'

distrib

archaeological record. The cultural and noncultural factors which
account for the initial deposition of wrecks (depositional transforms)
and the later decomposition of those wreck sites (archaeological
transforms) are discussed in great detail as they relate to the spatial

distributions revealed by this study.

A
those pec
However,
mention t
the finan
Sea Grant
and Recre
Monies re
asPects.
Charles c
Dr- Harry
thIOllghou
Bill Lovi
Prescribe
thanks go
Beth Hand
for final
£8110" co
not only ;

to the lIla:

ACKNOWLEDGMENTS

A few lines on this page are hardly enough to adequately thank
those people who have helped me in the completion of this project.
However, since this is the public format I am limited to I must briefly
‘mention the individuals who have given the most towards this end. On
the financial end of the matter-thanks are extended to the Michigan
Sea Grant Program.and to Dr, Donald Holecek of the Department of Parks
and Recreation.Resources at Michigan State University for research
monies relating to the study of Great Lakes shipwrecks from several
aspects. As for guidance-my committee composed of my chairman Dr.
Charles Cleland and members Dr. William.Lovis, Dr. Moreau Maxwell, and
Dr. Harry Raulet all provided me with useful comments and suggestions
throughout my graduate career. A very special thanks is reserved for
Bill Lovis for his intellectual and moral support far beyond his
prescribed academic duties. 0n the production side of this venture--
thanks goes out to my ex~wife Pam.Zwer for many house of typing, to
Beth Handrick for more hours of typing and help, and to Bonnie Graham
for final editing. Finally, I would like to acknowledge the help of my
fellow collegues at the MSU Museum who provided friendship and support
not only for this project but throughout the years. To those peoole and

to the many others who contributed to the cause--thanks!

ii

LIST OF 1

LIST OF 1

Chapter

INT

1.

‘i ‘§ 1 1“

PH?

II.

HIE

III.

E]\&

11E

.
m

V.
I

HHHHHU.HHHH.H-H.H.

TABLE OF CONTENTS

LIST OF TABIIES O O O O O 0 O O O O O O O 0

LIST OF FIGURES . . . . . . . . . . . . .

Chapter

I.

II.

INTRODUCTION . . . . . .

Problem Statement . . . . . . . .
Shipwrecks Defined . . . . . . . .

Shipwrecks and Underwater Archaeology

Previous Research . . . . . . . .

Sources of Data . . . . . . . . . . .

PHYSICAL ENVIRONMENT OF LAKE SUPERIOR

Physical Geography and Navigation
weather and Navigation . . . . . .
Environment and Vessel Loss . .

III.’ HISTORICAL BACKGROUND . . . . .

IV.

Early Development and Transportation .

The Iron Industry . . . . . . . .
Navigation and Industrial Expansion
The Grain Trade . . . . . . . . .
The Coal Trade . . . . . . . . . .

Historical Summary and Comparisons .

THE FORMATIVE PROCESS . . . . . .

Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis . . .
Hypotheses 10 and 11 .w. . . .
Hypotheses 12 and 13 . . . . .

‘OQNO\U\bWN-—

Hypothesis 14 . . . . . . . . . . .

Hypothesis 15 . . . . . . . . . .

iii

vi

ix

(AI-'00"—

~—

16

16
21
24

30

31
34
50
61
71
74

82

90
91
91
92
93
94
94
95
95
96
97
98
98

Chapter

HEI

1!. I. .l‘ I-

Chapter

Type B Archaeological Transform Hypothesis . . .

Hypotheses 16-21 . . . . . . . . . . . . .

v 0 “momma O O O O O O O O O

Shipwreck Attributes Defined .
Abstraction of Attributes from Historical Sources

Quantification of Attributes
Statistical Evaluation . . .
Poisson Distribution . .

Kohmogorov-Smirnov Test of Goodness

Contingency Chi Square .
KxZ Chi Square . . . . .
Interpretation of Results

VI. HYPOTHESIS TESTING .

Introduction . . . . . . . .
Hypotheses . . . . . . . . .

Type A Depositional Transform.Hypotheses

Hypothesis 1 . . . . . .
Test for Hypothesis 1

Hypothesis 1 Summary and Results .

Hypothesis 2 . . . . . .
Test for Hypothesis 2

Hypothesis 2 Summary and Results .

Hypothesis 3 . . . . . .
Hypothesis 4 . . . . .
Test for Hypothesis 4
Hypothesis 4 - Summary
Hypothesis 5 . . . . .
Test for Hypothesis 5
Hypothesis 5 - Summary
Hypothesis 6 . . . . . .
Test for Hypothesis 6
Hypothesis 6 - Summary
Hypothesis 7 . . . . . .
Test for Hypothesis 7
Hypothesis 7 - Summary
Hypothesis 8 . . . . . .
Test for Hypothesis 8
Hypothesis 8 - Summary
Hypothesis 9 . . . . . .
Test for Hypothesis 9
Hypothesis 9 - Results
Hypothesis 10 . . . . . .
Test for Hypothesis 10

Hypothesis 10 - Summary and Results

Hypothesis 11 . . . . . .
Test for Hypothesis 11

Hypothesis 11 - Summary and Results

and Results
and Results
and Results

and Results

and Results

and Summary

iv

of Fit

99
100

104

104
105
107
107
108
115
116
117
118

119

119
126
127
127
127
132
132
133
138
138
139
139
140
143
143
144
144
144
146
147
148
150
150
151
151
153
153
154
156
156
158
159
159
161

Chapter

Hypothesis 12 . . . . . . . . . . .
Test for Hypothesis 12 . . . . . . .

Hypothesis 12 - Summary and Results

Hypothesis 13 . . . . .
Test for Hypothesis 13 . . . . . . .

Hypothesis 13 - Summary and Results

Hypothesis 14 . . .
Test for Hypothesis 14 . . . . . . .

Hypothesis 14 - Summary and Results

Hypothesis 15 . . . .
Test for Hypothesis 15 .

Hypothesis 15 - Summary and Results

Hypothesis 16 . . . .
Test for Hypothesis 16 .

Hypothesis 16 - Summary and Results

Hypothesis 17 . . . .
Test for Hypothesis 17 .

Hypothesis 18.

Test for Hypothesis 18 .

Hypothesis 18 - Summary and Results

Hypothesis 19 .

Test for Hypothesis 19 . .
Hypothesis 19 - Summary and Results

Hypothesis 20 . . . . . . .
Test for Hypothesis 20 . . . . . . .

Hypothesis 20 - Summary and Results

Hypothesis 21 . . .
Test for Hypothesis 21 .

Hypothesis 21 - Summary and Results

Summary of Results . . .

Shipwreck Location
Commodity Affiliation

Vessel Type . . . . .
Loss Type . . .

Salvaged/Nonsalvaged

Other Variables . .

VII.

Interpretation of Results
Type A: Depositional Transforms

INTERPRETATION AND CONCLUSIONS .

Type B: Archaeological Transforms
Conclusion .

APPENDIX A . .

APPENDIX B - VESSEL DATA . .

”PENDIX C O O 0

REFERENCES CITED .

161
162
164
164
165
167
167
167
168
169
169
170
171
171
172
172
172
173
173
174
174
175
179
179
179
180
181
181
182
182
182
183
184
184
185
185

187
187
188
210
214
220
226
239

252

Ial

Table

“NGUIJ-‘wN—t
o

10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.

LIST OF TABLES

Lake Superior Harbors . . . . . . . . . . . . . .
Freight rates . . . . . . . . . . . . . . . . . .
Lake Superior Iron Ranges . . . . . . . . . . . .
Development of Ore Shipments . . . . . . . .
Expansion of locks at Sault Ste. Marie . . . .

Navigational nmprovements . . . ..... . . . .
Development of Great Lakes bulk carriers
Harbors of Lake Superior . . . . . . . . . . .

Lake Superior ports . . . . . . . . . . . . . . . .

Type A and Type B Factors . . . . . . . . . . . .
All commodities: total and salvage combined . . .
Poisson results for all Lake Superior shipwrecks
Iron: total and salvage combined . . . . . . . .
Iron: salvage and total losses combined . . . .

Grain: total and salvage combined . . . . . . . .
Grain: salvage and total losses combined . .
Coal: total and salvage combined . . . . .

Coal: salvage and total losses combined . . .
Iron 1855-1879 . . . . . . . . . . . . . . .
Iron 1880-1904 '. . . . . . ..... . . . .
Iron 1905-1929 . . . . . . . . . . . . . . .
Grain 1855-1879 . . . . . . . . . . . . . . .
Grain 1880-1904 . . . . . . . . . . . . . . .

Grain 1905-1929 . . . . . . . . . . . .
Coal 1855-1879 . . . . . . . . . . . .
Coal 1880-1904 . . . . . . . . . . . .
Coal 1905-1929 . . . . . . . . . . .
Iron, grain, and coal combined 1855-1879
Iron, grain, and coal combined 1880-1904

Iron, grain, and coal combined 1905-1929 . . . .

Summary of results . . . . . . . . . . . . . . .
Length of navigational season . . . . . . . . . .
Results . . . . . . . . . . . . . . .

Results of x test . . . . . . . . . . . . . . .
Marquette . . . . . . . . . . . . . . . . . . . .
Ashland . . . . . . . . . . . . . . . . . . . .
Two Harbors . . . . . .

Duluth/Superior . . . . . . . . . . . . . . . . . .
Port Ar thur . . O C O O O O O O C O O C . O O O O 0
Results . . C O O O C O O O O O C O O O O O O O O 0

Results 0 O O O O O O O O I O O O O O O O O O O O
Resu1t8 O O O O O O O O O O O O O O O O O O O O 0

vi

18

37

44

48

50

51

55

59

72

87
128
129
129
130
130
131
131
132
134
134
134
134
135
135
135
135
136
136
136
137
137
141
142
143
145
145
145
146
146
146
149
152

mmwwav

.mmummffm17

Table

43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
Ar]
A-2
A-3
A-4
A-5
B-l
B-2
B-3
B-4
B-5
B-6
C-1a
C-lb
C-1c
C-Id

Results . .
Grounding .
Foundering
Collision .
Fire . . . . . . . . . . . .
Results . . . . . . . . . . .
Results . . . . . .
Results . . . . .
Schooner . . . .
Schooner-barge .
Wooden steamer .
Steel steamer . . . . . . . .
Miscellaneous vessels . . . .

Results . . . . . . . . . . . .

Results . .

Traffic and loss . . . . . . .

Results . . . . . .
Results . . . . . . .
Iron - total loss . . . . .
Iron - salvaged . . . .
Grain - total loss . . . .
Grain - salvaged . .

Coal - total loss . . . . .
Coal - salvaged . . .
All commodities - total loss

All commodities - salvaged . . .

Results . . . . . . . . . . .
Results . . . . . . . . . . .
Results . . . . . . . . . . .
Percentage of vessels lost in
Iron - storms . . . . . . . .
Grain - storms . . . . . . .
Coal - storms . . . . . . . .
All commodities - storms . .
Summary . . . . . . . . . . .

Frequency of losses by industry in five
Opening and closing of navigation . . . . . . . .
Upper Peninsula forges and furnaces . .

Iron ore shipments by range .

Grain and flour shipments--Lake

Coal Shipments--Lake Superior
Iron Ore - Total Losses . . .
Iron Ore - Salvaged . . . . .
Grain - Total Losses .

Grain - Salvaged . . . . . .
Coal - Total Losses . . . . .
Coal - Salvaged . . . . . . .
All commodities - salvage and
Iron - salvage and total loss

October and November

0

O O O C 0

zones 0 . .

Superior 1870-1911

total loss combined

combined

Grain - salvage and total loss combined .

Coal - salvage and total loss

combined

vii

155
156
157
157
157
158
160
163
165
165
165
166
166
166
168
170
171
173
175
176
176
176
176
177
177
177
178
180
181
196
198
198
198
199
199
206
220
221
222
224
225
226
230
232
234
235
237
239
239
240
240

Table

C-Za 1E
C-2b It
C-Zc 11
C-Zd 1!
C-Ze :
C-Zf ll
C-Zg 1
C-Zh 1‘
C-Zi I
C-6a ‘1
C-6b A
C-6c T
C-6d D
C-6e P
C-lOa G
C-10b I
C-lOc C
C-lOd 1
C-13a s
C~ 13b 5
C~13c y
C-13d 5
C-13e 1
C‘ISa 1
C-19a
C-19b
C-lgc ‘
C‘Igd ‘
C-19e ‘
C‘19f 1
C‘19g
C~19h ‘
CODCIus
CencluS
COMIUS
COncluS

Table

C-2a
C-2b
C-2c
C-2d
C-2e
C-2f
C-2g
C-Zh
C-2i
C-6a
C-6b
C-6c
C-6d
C-6e
C-10a
C-10b
C-lOc
C-lOd
C-13a
C-13b
C-13c
C-13d
C-13e
C-15a
C-19a
C-19b
C-19c
C-19d
C-19e
C-19f
C-19g
C-19h

Conclusion.
Conclusion.
Conclusion.
Conclusion.

1855-1879 -
1855-1879 -
1855-1879 -
1880-1904 -
1880-1904 -
1880-1904 -
1905-1929 -
1905-1929 -
1905-1929 -
Marquette .

Iron
Grain
Coal
Iron
Grain
Coal
Iron
Grain
Coal

Ashland ......

Two Harbors

Duluth/Superior .

Port.Arthur

Grounding . . ......

Foundering
Collision .
Fire

Schooner . .

Schooner-barge .
Wooden steamer . . .
Steel steamer . .

'Miscellaneous vessels

Vessel capacity . .

Total loss

Iron - salvage . .

Grain - total loss

Grain - salvage .
Coal - total loss . . .
Coal - salvage

All commodities - total loss
All commodities - salvage .
Iron - stormw .
Grain - storms
Coal - storms . . .
All comodities - sto . . .

viii

O
O

O O O O O O
O

O O O O O O O

240
241
241
241
241
242
242
242
242
243
243
243
243
244
244
244
245
245
245
245
246
246
246
247
247
248
248
248
248
249
249
249
250
250
250
251

Fig

LIST OF FIGURES

Figure

1. Lake Superior harbors . . . . . . . . . . . . .

2. Lake Superior transportation routes . . . . . . . .

3. Iron ore production and transportation for the Lake
Superior Region 1855 - 1930 . . . . . . . . . . . .

4. Iron range locations . . . . . . . . . . . . . .

5. Freight rates for iron ore 1855 - 1928, Marquette to
Lower Lake Ports . . . . . . . . . . . . . . .

6. Iron, grain, coal tonnage on Lake Superior 1875-1910

7. Shipwrecks and the formative process . . . . . .

8. Variables in the formative process . . . . . .

9. Salvage and the formative process . . . . .

10. Poisson grid system . . . . . . . . . . .

11. Variables under consideration . . . . . . . . .

12. Variables affecting iron, grain, coal vessels . .

13. Iron-related shipwreck distributions . . . . . . .

14. Grain-related shipwreck distributions . . . .

15. Coal-related shipwreck distributions . . . .

ix

20
29

43
46

57
78
86
89
102
111
121
125
190
193
195

IESOU]
ical :
both '
logic
peepl
With“

Lakes

impo
SCie
r88:
arch
bro;
Com

the:

CHAPTER I

INTRODUCTION
Problem Statement

Shipwrecks in the Great Lakes constitute a major archaeological
resource that is currently poorly known and understood. As archaeolog-
ical sites, these wrecks offer potential for future investigations from
both regional and site-specific perspectives. And as with all archaeo-
logical phenomena, they reflect the culture and behavior of a past
people; a study of whom can enrich our understanding of culture process
within an anthropological perspective. The development of the Great
Lakes region and the subsequent industrialization during the nineteenth
century has significance for a number of academic disciplines, including
anthropology. Economic change, decision making, and development are
important areas of study to anthropology, as they are to all social
sciences. The shipping industry on the Great Lakes is one by-product of
regional economic behavior that has been recorded in the present
archaeological record through the occurrence of shipwrecks. From a
broad anthropological approach, the study of these shipwrecks can
contribute to the understanding of such cultural phenomena as economic
change and the industrial process.

Despite the potential research importance of shipwrecks as
archaeological sites, there has yet to be an anthropological treatment
of them in the Great Lakes region. Likewise, there is currently no

theoretical framework around which future shipwreck studies can be

organized. Shipwrecks differ from.terrestria1 archaeological sites
because they reflect a set of variables that are linked to transporta-
tion rather than settlement. Additionally, underwater wreck sites have
been deposited in the archaeological record by different processes than
land sites. These differences suggest that shipwrecks should be viewed
within a theoretical/methodological framework that both.recognizes their
unique qualities and directs future research.toward the investigation of
specialized problems. Recent academic study of shipwrecks in the Great
Lakes has primarily focused only on the historical events surrounding
individual vessel loss and not on broader issues. This narrow treatment
fails to recognize the cultural context of vessels on their value to the
interpretation of social phenomena.

The purpose of this dissertation is to apply a broad regional
approach.to shipwreck.studies and to begin laying a foundation for the
future investigation of wreck sites within an anthropological framework.
Of the many questions of potential research interest, the focal point for
this study is the spatial distribution of shipwrecks. This topic is
particularly appropriate as an initial treatment of the shipwreck
phenomena because it has value for both theoretical and methodological
issues. For example, shipwrecks are reflective of a well-documented
past cultural system. Since the shipping industry was known to be
structured by trade routes, industrial development, and the physical
environment, it would then be expected that to some extent shipwrecks
would likewise be patterned. This dissertation is framed around the
hypothesis that since shipping in the Great Lakes is a culturally
patterned phenomena and nonrandom in occurrence, shipwrecks will likewise

be patterned and nonrandom in their distribution. The question of

patterned regularities in the shipwreck archaeological record is of major
importance because it will substantively affect the future treatment of
shipwrecks. If shipwrecks are found to be distributed in a recognizable
pattern, then future excavations, sampling strategies, etc. can take

this into account and proceed accordingly. The type and extent of
patterning might also be viewed as a reflection of the cultural system
that shaped the development of the shipping industry; if so, patterning
has potential to reveal information on culture process that is not
recorded in existing documents.

As a predictive tool, the analysis of shipwreck locations may
also lead toward pattern recognition that can be applied to other
situations in order to locate unknown wreck sites or shipwreck.clusters.
For example, if this study reveals that vessels are lost in a specific
pattern in relation to particular ports or geographic features, than
similar locations can be sought elsewhere for like patterns. Although
less important for site specific studies, this question is of greatest
interest to those studies dealing with a broad regional approach to
shipwrecks. Future studies involving Great Lakes shipwrecks would hope-
fully be based upon the degree and type of spatial patterning found in
this study. Because shipwreck.studies of any type are new to Great Lakes
archaeology, one item of high priority is the location of sites and an
.understanding of the factors contributing to vessel loss and distribution.
Spatial analysis is a necessary first step toward generating research
interest in this little known area.

The study of the spatial patterning of shipwrecks involves much
more than just locational information. Rather, it requires an investi-

gation of a wide array of factors that contribute to vessel loss and

decomposition. For this reason, the study of the formative process of
the shipwreck archaeological record is also needed to fully understand
both.the patterns of distribution as well as the causal variables for
these patterns. In relation to this study, pattern recognition is not
an end in and of itself but is one means by which a theoretical base can
be initiated for the investigation of shipwreck.phenomena.

For the purpose of this study, a systems approach will be used
to order the data and to isolate those key variables of spatial pattern-
ing. Through.a series of testable hypotheses, the relationships between
these variables will be investigated and the results applied to an
interpretation of spatial distributions.

In order to condense the problem.into a manageable design, the
geographic area selected for study has been narrowed to the Lake
Superior region. This area was chosen because it was the last of the
Great Lakes to be settled and is therefore historically documented in
greater detail than the other Lakes. The time period considered here
begins with the opening of the St. Mary's Falls Ship Canal in 1855 and
ends with the second decade of the twentieth century. The year 1930.was
selected as a terminal point because few shipwrecks occurred after this
date because of modern advances in navigation, communication, and
vessel design. During this 75 year study period, the shipping industry
concentrated on the movement of four primary commodities: iron ore,
grain, coal, and lumber. The first three were of importance throughout
a large portion of this period, while the last achieved significance in
the post-1890 period. Therefore, to facilitate comparisons between
industries, only the iron, grain, and coal commodities will be dealt with

in the model.

Using these three industries, this study will determine if
vessels related to them are spatially patterned and if so, determine the
contributing factors to this distribution. Based on the results of
hypothesis testing, it may be found that some industry-related vessels
exhibit patterns of loss different from the others. If so, then
potential alternative explanations will be proposed for these patterns
in light of the relationships between other variables tested in the
analysis. It is hoped that the broader issues addressed to spatial
patterning will have significance beyond this single geographic area and
the time period. Because this study addresses Lake Superior industries,
it must necessarily be oriented toward economics and spatial distribu-
tions. And an underlying goal of this study is to demonstrate the
utility of a broad systems approach to a topic that has been investi-
gated in the past from a historical/particularist perspective. Ship-
wrecks on the Great Lakes are a potential source of information on broad
social issues such as economic development and technological change. In
the past, these issues have been ignored by historians in favor of more
particularistic treatments of individual vessels and their conditions of
loss. As a result, shipwrecks are presently viewed in the same manner
as historic buildings-—that is, that they are valued only in terms of
their unique architectural features and not in relation to the informa-
tion they can provide toward an understanding of culture process and
change.

In this study, the specific details of individual vessels are
used to solve particular problems related to the study of spatial
distributions. Such variables as vessel type, condition of loss, date

of loss, industry affiliation, etc. will be analyzed as components of an

interrelated system. The shipping industry on Lake Superior was part of
an industrial system that significantly changed the character of American
culture in the Great Lakes region. As elements of this system, such
factors as economic decision making, transportation technology,

regional development, and the physical environment played a major role

in shaping the transportation industry.

A locational study of shipping and shipwrecks from such a
perspective requires an understanding of the processes that form the
archaeological record. The system within which the shipping industry
operated provides a setting for the understanding of shipwrecks in an
archaeological context. A systems approach to wreck phenomena recognizes
the cultural context of vessels and places specific characteristics such

as spatial distributions within this broader system.

Shipwrecks Defined

 

The shipping industry in the Great Lakes was a complex system of
transportation responsible for the movement of vast numbers of people,
raw.materials, and finished goods. This transportation system not only
represented the economic behavior of the time but also became a develop-
mental force that influenced the direction and future economic expansion
of the region.

The Lake Superior region provided a relatively undeveloped geo-
graphic area on which the shipping industry could operate. Before the
opening of the St. Mary's Canal at Sault Ste. Marie in 1855, the "shipping
industry" consisted of the movement of small amounts of goods between the
settlements still involved in the final years of the fur trade era. With
the discovery of mineral and timber resources in the Lake Superior Region

at a time when America was feeling the demands of the Industrial

Rev:

cla:

the

tin

net

1201

C31

C01

8C

th

re

th

51'

Revolution, the region became of immediate interest to the emerging
class of wealthy eastern entrepreneurs. The subsequent development of
the region, first in mineral resources and then in agricultural and
timber commodities, resulted in the rapid expansion of transportation
networks, including the railroad and shipping industries. water trans-
portation in particular was easily adaptable to the bulk movement of
cargoes and soon became the dominant mode of transportation for such
commodities as iron ore, grain, and coal.

Throughout the years of Great Lakes navigation, a great many
accidents occurred involving thousands of vessels. The majority of
these mishaps resulted only in minor damage and after a brief period of
repair, the vessel returned to the water. Some vessels were lucky in
their careers on the lakes and survived many years of service without a
single mishap. Others were not nearly so fortunate and were "jinxed"
with one accident after another throughout their lifetimes. Some of
these accidents resulted in the partial or total loss of vessels, their
cargoes, and their crews. The loss of life and property were of major
interest to the many small communities of the region-and in particular
to those with ties to the shipping industry. News of accidents traveled
quickly and newspapers hungry for a story quickly turned mishaps into
front page stories. Thus, shipwrecks became an unfortunate but
integral part of society and through.written and oral accounts, they
became a part of the local history and tradition.

On first inspection, the meaning of the term "shipwrecks" seems
quite obvious and of little need of further elaboration. However, in
actuality this term has been used to represent a wide variety of mishaps

varying from minor accidents to large and costly disasters. Since this

study deals ultimately with.the shipwreck archaeological record, it is
necessary to draw finer distinctions between the severity of loss
experienced by vessels. In particular, it is necessary to separate
vessel mishaps into one of three categories: 1) minor accidents where
the vessel(s) involved remained afloat and did not require extraordinary
assistance to make port; 2) accidents that incapacitated a vessel for a
period of time (days to years) and that required special assistance for
the vessel to make port; and 3) accidents resulting in total loss of the
vessel so that even special assistance could not put it back.into
service. These three distinctions are necessary to fully understand the
formative process of the archaeological record. For example, minor
accidents would contribute little or nothing to the archaeological
record and no traces of these actions remain today. On the other hand,
accidents that resulted in total loss of the vessel comprise a major
portion of today's shipwreck record. The second class of accidents is
likewise of importance because it represents those vessels that were
once a part of the archaeological record but were removed through.such
human actions as salvage. Although this type of accident is not repre—
sented today in the record, it is important to the understanding of past
cultural actions as well as the process by which the present shipwreck
record is formed.

This study will consider only vessels of these last two types
(i.e., salvaged, total losses) because of the relative lack of accurate
information on minor accidents. Since major accidents were reported
widely in newspapers of the period, better information is currently
available on these diasters than on the less newsworthy minor mishaps.

This is not to say that minor accidents are not worthy of further

investigation, but only that they cannot be dealt with in this study

because of a lack of detailed information.

Shipwrecks and Underwater Archaeology

Shipwrecks have been a part of the human experiences as long as
watercraft have sailed the seas and waterways of the world. For more
than nine thousand years, vessels have moved peoples, products, and
ideas across space and between cultures (Bass 1972:12). It is not
unusual then that water transportation evolved rapidly as an adaptation
to a world that is three fourths covered by water. Over time, water-
craft have proved to be the single most efficient means of transportation
available. However, as long as ships have sailed the oceans, storms
have destroyed those vessels. Although unfortunate for the crews of
those ships, without those shipwrecks we would know very little today of
these early craft and of the cultures that produced them.

Underwater archaeology had its beginnings in the early 1960s with
the excavation of a Bronze Age ship in the Mediterranean (Bass 1972:9).
Since then, underwater archaeology has grown rapidly as a field of study
with its practitioners coming from a number of different academic
disciplines. The present orientation of this field is very similar to
that of European schools of archaeology whereby history, rather than
anthropology, is stressed. This in part has been due to the internation-
al character of the field, with.many of the first underwater excavations
conducted either by European or Scandanavian teams of researchers. Like-
wise, the involvement of classical archaeologists with their historical
or art-historical approach.has added further to the stressing of this

nonanthropological orientation to underwater archaeology.

10

The newness of underwater sites to many archaeologists has led
also to an overemphasis on description and classification of marine
artifacts and vessels. Those few anthropologically trained archaeolo-
gists who have entered the underwater realm have been reluctant--or
unable-to address broad cultural questions using shipwreck data. In
many ways, underwater archaeology today resembles American prehistoric
archaeology of the mid to late nineteenth century. According to Willey
and Sabloff (1974:42-87), the Classificatory-Descriptive Period (1840-
1914) was characterized by the "description of archaeological materials,
especially architecture and monuments, and the rudimentary classification
of these." In many ways, this statement also accurately describes the
large majority of current underwater archaeological investigations.

The combination of a European historical view toward archaeology
and a fixation (even by many anthropologists) on the architectural
description of vessels has led to a disinterest in underwater cultural
phenomena by most scientific anthropological archaeologists. This is
extremely unfortunate because these phenomena have tremendous potential
to the study of cultural process and culture history. Until this
situation changes, the underwater resources currently available for
study will not be utilized to their full potential and may perhaps be
destroyed by well-intentioned but poorly trained archaeologists.

In recent years, Great Lakes shipwrecks have received a great
deal of attention in the popular literature but have for the most part
been largely ignored by the archaeological community. This has been due
to a number of factors, including: 1) lack of water-based training by
the majority of land archaeologists; 2) the fact that most underwater

archaeologists have their training in marine rather than fresh water

11

areas; 3) lack of academic programs in Great Lakes region universities;
4) the lack of conservation facilities for handling large scale under-
water excavations; and 5) the reluctance of Great Lakes state govern-
ments (i.e., history divisions, museums, etc.) to become involved with
a new area of study. These and other reasons have contributed to a
very poor climate for the study of Great Lakes shipwrecks from an
archaeological, and particularly anthropological, viewpoint. It is not
surprising then that problem—oriented research in the region has not

been undertaken to this point in time.

Previous Research

 

Outside of the fields of history and marine history, very
little academic research has been directed toward shipwreck studies in
the Great Lakes, partly because of the nature of underwater archaeology
and its relatively expensive methods of field excavation. Despite the
lack of archaeological emphasis on shipwrecks, several historical
treatments of this phenomena are noteworthy. wright‘s (1972) study in
particular is unique because it was directed toward a broad approach to
wrecks rather than the usual particularistic case study of the more
"important" vessels. His study was commissioned by the Michigan
Department of Natural Resources and was designed to be an inventory of
vessels thought to be lost within Michigan waters. But the ultimate
purpose of the study was to detect geographic areas with particularly
high concentrations of wrecked ships and to proVide an overall view of
the shipwreck resource in the state. Based on numerous primary and
secondary sources, this study succeeded in inventorying a total of
1,139 vessels believed to have been lost in state waters. Of these,

175 were believed lost in Lake Superior (wright 1972:6). For each

12

vessel, characteristics such as general location, vessel name, vessel
type, type of loss, date of loss, and cargo type were recorded when
available. This information was then computerized and cross-tabulations
were run on possible variable combinations. Unfortunately, the study
progressed no further and no effort was made to analyze the results of
the_project nor to place the vessels into an overall developmental
framework. Therefore, the Wright study was primarily descriptive; it
presented information on shipwrecks but did not relate the findings to
the development of the Upper Great Lakes region.

Wright (1972) also attempted to deal with the problem of
location by dividing Michigan waters into 18 arbitrary zones. Locations
of vessel loss were then grouped into these zones and could be corre-
lated with other factors to form.a coastal shipwreck profile, which
consisted of the number of vessels lost in an area along with data on
the relative vessel types, cargoes, types of loss, etc. Wright hoped
that this locational study would determine general patterns of loss for
specific areas. While to some extent this was accomplished, there were
a number of problems associated with this approach. For example, the
location zones dealt only with.the Michigan waters of Lake Superior,
while the study vessels had actually been a part of the total Lake
Superior transportation system. By looking only athichigan waters,
wright could have obtained a very biased representation of Lake
Superior shipwrecks. In addition, the location zones were entirely
arbitrary and bore no resemblance to the cultural boundaries of the
shipping system. Without an understanding of the major ports, trans-
portation networks, and natural resource locations, it would be

difficult to arrive at culturally meaningful locational zones. For

13

these reasons, Wright's treatment of locations was necessarily limiting,
although it was an excellent early attempt to deal with.the question of
spatial distribution of shipwrecks.

Other than this single study, no other research projects have
been conducted on Lake Superior shipwrecks from anything other than a
historically descriptive standpoint. Relevant historical studies will

be discussed in the following section.

Sources of Data

 

This research.project will incorporate a wide range of histori-
cal sources, both primary and secondary, into the investigation of
shipping and shipwrecks in the Great Lakes, and will focus on applying
the large body of historic data available on this topic to questions
posed within a broader anthropological framework. Historical treat-
ments of the shipping industry are filled with an abundance of data on
vessels, vessel patterns, and shipwreck locations. What is proposed in
this study is to utilize these extant historical sources as a data base
for testing broader based anthropological questions. There are
essentially two general categories of sources used in this investiga-
tion: 1) those sources providing information on the shipping and
economic history of the Lake Superior region; and 2) those sources
dealing specifically with shipwrecks and providing information about
their origins and locations. Of these two types of sources, the former
is by far the largest and most complete. This is because of the impor-
tance of shipping and shipping-related industry to the Lake Superior
region and the extensive collections of materials available on these
subjects. A wide range of primary source materials are available in-

cluding commodity records, shipping records, and government statistical

14

abstractsr Additionally, several scholarly studies on selected topics
have been published that provide an excellent base of secondary sources
from which to draw. A combination of both primary and secondary
sources will be used in this research project to outline the shipping
industry of the period 1855-1920. In particular, such sources as
IMansfield (1899), Barry (1973), Hatcher (1950), Nute (1944),

Williamson (1977), and Wright (1972) provide detailed accounts of
specific vessels, industries, or chronological periods of relevance to
this study.

In contrast to the abundance of materials available on shipping
and related industries, information on individual shipwrecks is much
scarcer and difficult to obtain. The major primary source on ship-
wrecks in Lake Superior are the Annual Reports of the U.S. Life-Saving_
Service for the years 1877-1914. These reports are yearly summaries of
all services provided to distressed vessels and include information on
specific vessels, their cargoes, locations and damage. The Life-Saving
Service divides the Great Lakes to encompass three districts--No. 10
(Lake Ontario and Lake Erie), No. 11 (Lake Huron and Lake Superior),
and No. 12 (Lake Michigan). Although these records often include
reports on lost vessels, they are often incomplete.

Other primary sources include newspapers, trade journals, and
vessel registers, and each contain sizable amounts of data of varying
quality and accuracy. There are a large number of secondary sources on
shipwrecks and these too are a mixture of scholarly and sensationalized
historical accounts of lost vessels. .For Lake Superior, the single

most significant source is Wolff's (1979) The Shipwrecks of Lake

 

Superior. This is the most complete and well-documented treatment of

15

wrecks in the Great Lakes and uses an historical case study approach.
to examine well over 1,000 accidents. Wolff‘s 23 year study of Lake
Superior shipwrecks includes a wide.variety of information drawn from
primary sources such.as those previous mentioned. Because of its high
degree of scholarship, this study will serve as the major source of
shipwreck data for this research project. Other secondary sources such
as Heden (1966) and Winkelman (1971) will be used as supplementary data
whenever needed. When combined, these sources offer the most reliable
and complete data presently available on the subject.

Therefore, the main emphasis of this dissertation will be the
application of existing historical data to questions relevant to the
study of spatial distributions of wrecks. Although this approach
relies on the quality of available sources, this does not appear to be
limiting since they are complete enough to provide the data needed to
test hypotheses posed in this investigation. Those hypotheses and the
model around which they are organized will be discussed in detail in

following chapters.

CHAPTER II

PHYSICAL ENVIRONMENT OF LAKE SUPERIOR

The physical environment of the Lake Superior region had a
major effect on the type, extent, and rapidity of its settlement and
development. The location of natural resources, population centers,
and transportation lines encouraged development of certain sectors of
the region over others. This chapter outlines the physical geography
of Lake Superior and its coastline, illustrating the prominent features

that affected navigation and shipping.

Physical Geography and Navigation

With a surface area of approximately 32,000 square miles, Lake
Superior is the largest body of fresh water in the U.S. Its greatest
length from east to west is about 350 miles, while the point of great-
est width is roughly half as long. The altitude of the lake surface
averages 600 feet above sea level, 20 feet higher than Lakes Michigan
or Huron at 580 feet (Sommers 1977:43). Until the construction of a
lock system at Sault Ste. Marie in 1855, this difference in levels
created a barrier to the direct access of the lake. During the last
century, the level of Lake Superior has remained more stable than the
other Great Lakes, with a maximum variation of only 2.6 feet. These
yearly variations are due to a number of factors, including temperature
and rainfall and may have had an impact on navigation by making some

harbors and ports less accessible during lowbwater periods.

16

17

Lake Superior is also the deepest of the lakes; its maximum
depth of 1333 feet is found in the eastern portion of the lake. While
variations in depth (over 30 feet) would have little direct impact on
navigation, they would definitely play a role in the salvage and
retrieval of lost vessels and cargo.
The waters of Lake Superior are clearer and colder than the
other Great Lakes. Average water temperature at Marquette, for
example, ranges from a high of 60° F in August to a low of 34° F from
December through.March (Mansfield 1899:46). During the winter months,
it is not unusual for the lake to freeze over, either totally or
partially. Even in the summer months the water temperature under the
surface is greatly reduced for each.20-foot increment of depth, so that
below 200 feet the temperature remains a constant 39°F. These tempera-
tures, along with the fresh.water, combine to produce conditions
perfect for preservation of organic materials.
Lake Superior harbors are particularly important to lake
navigation by providing terminal points for docking as well as a safe
refuge in case of storms. An early traveler on the lake commented that
The natural harbors of this lake are not numerous, but on account
of its extent and depth it affords an abundance of searoom, and
is consequently one of the safest of the great lakes to navigate.
The only trouble is that it is subject to severe storms which.
arise very suddenly. Often have I floated on its sleeping bosom
in my canoe at noonday and watched the butterfly sporting in the
sunbeams; and at the sunset hour of the same day have stood in
perfect terror upon the rocky shore gazing upon the mighty
billows careening onward as if mad with a wild delight, while a
wailing song, mingled with the 'trampling surf,‘ would ascend to
the gloomy sky (Mansfield 1899).

This quotation graphically illustrates the relative lack of harbors as

well as the urgency of finding refuge from.storms. Following is a

listing of the principal harbors found on Lake Superior.

18

Table 1. Lake Superior Harbors

Sault Ste. Marie, ME, ONT Eagle Harbor Duluth/Superior, MN/WI
Grand Marais Ontanagon Two Harbors

Munising Ashland, WI Grand Marais, MN
JMarquette La Pointe, WI Port Arthur, ONT
L'Anse Bayfield, WI

Copper Harbor Portwing, WI

These harbors constitute the major geographic areas where a
number of factors combine to allow for port facilities. Such.elements
as water depth and shelter are essential for the construction and use
of docking facilities. From.the list above, it can be seen that on the
entire north shore of the lake there are only three harbors--Two
Harbors and Port Arthur-~compared to 14 (88 percent) on the south shore.
In an historical sense, this geographic discrepancy has had a major
effect on the settlement of the respective shores of the lake (see
Figure 1).

In times of storms, vessels on the lake would have attempted to
either "weather them out" or head for safe harbor. Because of the
relative lack of facilities for such a large geographic area, the next
best alternative would be to seek shelter wherever possible. Protected
areas are those locations that provide vessels some relief from the
winds and waves associated with the dangerous northwest storms of the
region. When a vessel is caught on the open lake in a storm too
violent to ride out, it seeks one of these sheltered areas, which are
predominately on the leeward side of points, peninsulas, and islands.

Navigational hazards on Lake Superior are difficult to define
since this depends on human behavior as well as geography. For
example, those same islands that provide shelter from storms are also

potentially hazardous to navigation under other circumstances, such as

\OGJNChUI-L‘UNH

10.
ll.
12.
l3.
14.
15.
16.

Figure l.

Sault Ste. Marie, MI, ONT
Grand Marais,.MI
Munising, ME
Marquette,.MI

L'Anse, ME

Copper Harbor, MI
Eagle Harbor, MI
Ontanagon, MI

Ashland, WI

La Pointe, WI
Bayfield, WI

Portwing, WI
Duluth/Superior, MN/WI
Two Harbors, MN

Grand Marais, MN

Port Arthur, ONT

Lake Superior Harbors

19

20

 

 

flljllJflHHmulJ

muzz Bad—(pm

u.u($ up... n
ba3(m 02.2231
0... 0
96 awkwacai‘c‘
by
a u
unZKa Q
Q \
0—1030.
.. v
O
. o .
O0.
‘0 wOS—uOs
. s 9.2.0
no
. . a . 3
Z . sun—239:.
. 0
§

06

G 02(d1w<

 

 

 

21

fog. Therefore, hazards are relative phenomena dependent upon a number
of factors that come together at a single point in time. Shoals found
along large portions of the coastline present a hazard to navigation
only when vessels stray outside of their designated navigational lanes.
In geographic areas where a course change in necessary (e.g., for
rounding a point, or avoiding an island), weather conditions are
hazardous only when they reduce visibility or maneuverability. Some-
times ships might actually seek out hazardous areas: intentional
grounding could prevent total loss to the vessel and crew when
foundering was the only other alternative; WOlff (1979) discusses
numerous instances when these intentional groundings became necessary.
So navigation on Lake Superior was affected by a wide array of both
environmental and cultural factors; shipping and the resultant loss of

vessels was not a function purely of climate or geography.

Weather and Navigation

 

Throughout the history of Great Lakes navigation, weather had
a major influence on the movement of ships and cargo across this broad
geographical area. weather can be viewed in two ways: as a commercially
restrictive force and as a randomizing element in the navigation
process. In terms of its restrictive or constraining feature, the
single element of greatest influence was the season of navigation on
the lakes. The length of the season had a major effect on the amount
of goods. the number of vessels, and the number of round trips possible
in a given year. Table 1, Appendix A, presents a listing of the length
of navigational seasons on the lakes for the years 1855-1930. These
dates represent the recognized opening of the Sault Canal and pertain

especially to the traffic from Lake Superior to southern ports.

22

During seasons of short duration, the traffic on the Lakes is
congested into a shorter period and vessels are anxious to sail at the
beginning and ending of the navigation season. In addition, in an
extremely short season with an early winter, economics often force
vessels into attempting a final round trip when in longer season years,
this would not be necessary.

The weather also influences traffic on the Lakes through storm
and fog patterns. Storms on the Lakes can approach those of the oceans
in both intensity and duration. But the Lakes present certain features
unique to this geographic area. Generally, most major storms on Lake
Superior occur during October and Nbvember, while the best weather is
in the months of July and August (Mansfield 1899:45; Townsend and
Ericson 1978:123-125). Although much more frequent in October and
Navember, storms may arise any time throughout the year. Before the
advent of ship-to—shore communications in the first decade of the
twentieth century, ships were more or less dependent upon the judgment
of the captain and crew for predicting shifts in weather. Even with
these communication facilities, many ships are caught in rapidly rising
storms at locations that offer no refuge. So it is a matter of chance
exactly when and where storms will interfere with.navigation and
possibly result in vessel loss. Most of the severe storms come from.the
north, as opposed to the prevailing winds that normally come from the
southwest or west (Mansfield 1899:47). This northerly shift of wind is
particularly dangerous because large fluctuations in water level
occur over the unrestricted water, sometimes resulting in seiches and
surges. These are particularly dangerous in shallow areas along the

coastline, such as in Keweenaw Bay.

23

As construction of steel--and 1arger--bulk carriers began and
as weather prediction became more scientific, the number of vessels lost
in storms began to decrease. With the advent of radar as standard
equipment in 1940, collisions and groundings were significantly
reduced. For example, Lake Superior collisions accounted for eight of
136 losses during the period 1835-1900, and from 1900-1943, this number
increased to 13 of 143. But with radar after 1943, no vessels have
been lost in collision (Wolff 1962:136). Fog was one major cause of
collision. In the Great Lakes, fog is most prevalent in the northern
regions, such as Lake Superior, where the water stays colder. It
occurs most often during a warm trend in the spring when the water
temperature lags behind the temperature of the surrounding land. By
summer, the water temperature has warmed to a point where fog is not a
problem (Townsend and Ericson 1978:125-126). Generally, fog is a
problem.within five nautical miles of the shoreline and does not pose
a major threat on the open lake. Beyond its visually limiting effect,
one characteristic of fog, is its abberation of sound. Ship and fog
signals on the Lakes can easily be misjudged so that the sound appears
to be coming from a different direction and distance than it actually
is (Mansfield 1899:50). Possibly for this reason few fog signals were
ever installed on Lake Superior. The Coast Guards Light list carries
this warning:

Fog signals depend on the transmission of sound through air. As
aids to navigation they have certain inherent defects that should
be considered. Sound travels through air in a variable and
unpredictable manner. IMariners are warned that fog signals can
never be implicitly relied upon (U.S. Coast Guard 1970),

(Holland 1972:202-203).

Collisions and groundings in fog remained a major problem on Lake

Superior until the 1940s.

24

Environment and Vessel Loss

 

Collisions were only one of four major causes of loss on the
Great Lakes; fire, founderings, and groundings were also major forms of
vessel loss that took a heavy toll. Fire was more common to certain
types of vessels than others. Early propellors, steamers, and tugs were
particularly susceptible to fire because their propulsion systems were
dependent on wood- or coal-fired boilers (Wright 1977:12). Naturally,
they presented a constant hazard, especially in stress situations.

The other disasters--founderings and groundings--were common to
all types of vessels but particularly to those sailing ships with a
greater dependency on the weather. For example, schooners, which were
generally small in size, were common on the lakes before the institution
of communication devices and many modern navigational aids. ‘Many
sailing ships were lost through foundering or grounding in the intense
storms of Lake Superior. With the advent of larger and safer vessels,
losses in storms were greatly reduced. Between 1920 and 1940, the
institution of separated navigation lanes and radar eventually eliminated
losses due to collisions (Walff 1962:137). In this same period,
founderings were also reduced. Ice build-up from winter storms,
shifting of cargo in storms, etc., all played a role in founderings.
Technological advancements in vessel design and cargo handling made the
modern bulk freighters very stable in the water. By the 1940s, ship
loss due to any type of disaster was an unusual occurrence. No single
factor was responsible for the improvement of vessel safety, rather it
was a combination of different but related phenomena that played a
'major role in this regard. Larger and more stable vessels, better

weather reporting and communication systems, radar, and well-defined

25

navigation lanes all helped eliminate major disasters on the Great
Lakes.

Until the 1920s, the actual navigational routes on the Lakes
were informally derived. At that time, sailing lanes separated for
upbound and downbound traffic became separated in order to minimize the
danger of collisions (Johnson 1948:118). Before 1920, navigational
routes were charted by individual vessels or companies dependent upon
the destination and stops a vessel had planned. Because of the natures
of cargoes and the specifics of each industry, there were major differ-
ences in navigational routes between shippers of different commodities.
Therefore, routes traveled by carriers of package freight were
different from are carriers because they had different markets to cover
and variations in points of origin and destination. An ore carrier
downbound from.Marquette would plan a direct route to Lake Erie to
minimize time and expense. A package freighter traveling the same
route might stop at numerous intermediate points to load and unload
cargo, covering different routes in the process. The same can be said
for vessels carrying other commodities like lumber, grain, salt, and
stone, all common to Lake Superior traffic.

Before the establishment of set navigational lanes, the judg-
ment of the ships' masters (along with some basic rules for passing,
etc.) were the only regulations imposed on vessel traffic. This pre-
sented few problema when the weather was clear and visibility good but
in storms or heavy fog, the danger of collision or grounding was very
real. The greatest dangers occurred where navigation was so restricted
that traffic concentrated in certain areas. The geography of Lake

Superior is such that at some points, maneuverability is restricted

26

because of channel depth.or navigational hazards. Five areas in
particular have this concentrating effect: 1) Duluth/Superior to the
Apostle Isle; 2) Ashland past the Apostle Islands; 3) Keweenaw
Peninsula; 4) Whitefish Point; and 5) Southern Whitefish Bay to Sault
Ste. Marie. In these areas, vessels might change course several times
to maintain a direct route across the lake. Also, ship masters seek to
minimize the distances they must travel because of the economics of
lake carriers. Time lost going too far out when rounding the Keweenaw
is lost money and broken schedules at the furnaces on Lake Erie. Some
concentration of traffic is therefore a result not only of geographic
constraints but also of maximizing economic behavior by vessel captains.
This congestion of traffic around harbors significantly increases the
probability of vessel loss due to collision or grounding.

Ships associated with the bulk cargo industries followed a
general pattern of navigation consistent with the developments of the
shipping industry as a whole. When the iron industry opened new
ranges for mining, new lines of transportation developed accordingly.
Within this major overall shipment pattern, individual ships varied
little in regard to the specific routes traveled on Lake Superior. The
downbound cargoes of pig iron or iron ore traveled routes designed to
maximize speed and to minimize distance in the journey to Lake Erie
ports. Vessels traveled straight courses (wind and weather permitting)
that took them across the open lake. Upbound vessels usually brought
cargoes of coal, general freight, and sometimes passengers as vessel
design permitted. These trips were normally made on similar routes that
brought the vessels slightly closer to shore on the upbound leg of the

journey. The navigational lanes that were eventually established

27

through federal regulations are very similar to the historical routes
established during the major period of ore shipment beginning in 1870.
Figure 2 illustrates the major modern navigational routes on Lake

Superior.

28

Figure 2. Lake Superior Transportation Routes

29

 

 

u.¢(¢< men
n :(m

  
      

1 H mm _m in 0

«~21 w~:h(pm

02.2231

fl
o

mphuaflv¢<S

 

OpZODO .

s
«B

 

‘0 uO<~ cOm

02(30

>3
sun-2::-

Oz<d:m(

\
32.... an

maOnu<1
O3—

 

 

CHAPTER III
HISTORICAL BACKGROUND

This chapter introduces a general history of the Lake Superior
shipping industry and discusses the specific development of the iron,
grain, and coal industries as they relate both to shipping and to the
development of the region. Because the shipping industry of Lake
Superior was so highly interrelated with the various commodity trades,
there.was a great deal of overlap between these two areas. Such factors
as technological change in vessel construction and navigational improve—
ments will also be discussed as they apply to the overall development
and evolution of the shipping industry.

This chapter is not meant to be a complete treatment of this
enormous topic, but rather is a synopsis of salient points relevant to
the problem under consideration. From this summary, general patterns
of chronological development and shipping history can be drawn that
will be used to provide the cultural context for examining shipwrecks.
The historical factors that affected production, transportation, and
distribution of various commodities also affected the characteristics
of resultant vessel losses. In order to understand the interrelation-
ships between factors influencing the shipping industry, a historical

perspective is required.

30

31

Early Development and Transportation

 

Early explorations into the Lake Superior region were an out-
growth of overall development and settlement of the Great Lakes in the
first half of the nineteenth century. Earlier experiences in the Lake
Superior basin, such as the fur trade and the War of 1812, succeeded in
bringing the extensive timber resources to public attention. In the
succeeding several decades after the war of 1812, the character of the
eastern United States underwent a rapid change because of industriali-
zation and the associated growth of urban areas. The Great Lakes were
minimally involved in the first wave of the industrial revolution, and
it was not until the appearance of lake steamers in 1818 and the opening
.of the Erie Canal in 1825 that states of our present Midwest, such as

Ohio and Michigan, were settled to any great degree (Bald 1954:152).

With the opening of the upper lakes to an eastern transportation
network, development accelerated rapidly in terms of both population
expansion and preindustrial commercial ventures. Before 1830, commerce
west of Detroit was primarily involved with the Indian trade, while the
trade east to Buffalo consisted of eastbound fish, furs, corn, and
whiskey and westbound cargoes of passengers and general merchandise
(Mansfield 1899:182). With.Buffalo as the major eastern port, it became
the early commercial center for Great Lakes traffic and development.

By 1832, more than half of all westbound immigrants arrived via Great
Lakes waterways. Immigrants from EurOpe and the northern states used
the Great Lakes transportation system almost exclusively (Mansfield
1899:183). Much of this western movement resulted from the investment
in the upper Great Lakes region by wealthy eastern land speculators and

from the corresponding development of industry and commerce along the

32

Lakes. During the period 1830 to 1840, the population of Michigan rose
from 31,639 to over 212,000, and by 1850 to just under 400,000. This
immense growth was concentrated in rural agricultural areas along major
transportation routes such as the lakes and major rivers.

Throughout this period, water transportation played the major
role in the distribution of people and goods across the region because
of its relative inexpensiveness compared to overland travel. Despite
its great importance, until the mid-18508 the water transportation
system on the upper lakes was still in its infancy in comparison to the
lower lakes. By 1840, only eight vessels were making the trip from
Buffalo to Chicago on a regular basis (Mansfield 1899:185).

Overland transportation was also on the rise at this time,
providing substantial competition for the passenger trade. Roads from
Detroit to Saginaw, St. Joseph, and Chicago were completed by the mid-
1830s and stage service was established along these lines (Parkins
1918:258-261). Before the railroads arrived, these inland roads served
as the only outlets for surplus agricultural products for much of
interior southern.Michigan. Detroit was the nucleus from which these
roads radiated and became the major marketplace as well as the supply
station for the state (Parkins 1918:262).

Beginning in the 1840s, the building of railroads in Michigan
became a major industrial venture. Immediately upon completion of
the Michigan Southern and the Michigan Central railroads to Chicago in
1852, the railroads became major competitors of the steamahips
companies for business. Having the advantage of year-round service,
they soon replaced water transportation as the leader in passenger

traffic (Mansfield 1899:191). This had a major effect on the shipping

33

industry in terms of the movement of people but had no effect on the
movement of bulk goods. Rates for freight remained considerably less
for water transportation than for any form of overland travel.

Even though Michigan saw rapid expansion in both population and
transportation, it remained an agriculturally oriented state before the
Civil War (Bald 1954:249). The period of 1830-1855 could therefore be
regarded as a preindustrial frontier of the industrial revolution which
was rapidly pushing westward from eastern centers. This frontier
followed transportation routes based on water and rail lines, and
development followed quickly in its wake.

In the eastern United States during the mid-nineteenth century,
the industrial revolution changed many industries through the intro-
duction of the factory system. The seemingly endless expansion of
railroad networks in both the east and the west also transformed the
timber and iron industries to meet increasing industrial demands. At
that time, iron production was centered in Pennsylvania where abundances
of hardwood forests, coal, and iron ore were in close enough proximity
to each other to make it a profitable undertaking. The hardwood forests
were cut and reduced to charcoal, which.produced a high quality,
malleable wrought iron. But the eastern forests did not last long and
by 1855, coal-smelted iron had for the first time surpassed charcoal
iron in percentage of total output (LaFayette 1977: ix). By 1850, the
demand for iron created a market situation that pushed the price of
finished iron into a steady rise that lasted throughout the century.
When A.W} Burt discovered iron in the Lake Superior region in September
of 1844, a number of entrepreneurs soon recognized the significance of

his find.

34

The Iron Industry

Just before the discovery of iron ore near Teal Lake (southwest
of Marquette), a company was forming to mine copper. When word of the
discovery became known, this company shifted its interest to iron and
in July of 1845, the Jackson.Mining Company was formed. In 1847, they
began mining operations and erected a Catalan forge. Destroyed by a
flood later that year, the forge was rebuilt in 1848 and continued an
unsuccessful commercial operation until 1850, when it was abandoned.

In 1849, two eastern industrialists organized another company interested
in iron production, naming it the Marquette Iron Company. This company
built a forge in 1850 and by the next year, produced iron at the rate

of about 20 tons per week (LaFayette 1977:2).

The Cleveland Iron Mining Company was founded in 1847 to buy
and explore irbn ore lands. In May of 1853, the Marquette Iron Company
sold out to the Cleveland Iron Mining Company (Bald 1954:241). Like the
other companies before it, the Cleveland Iron Company was also involved
with the production of iron in forges as well as with limited shipment
of ore. The first shipment of iron ore from.Lake Superior occurred in
July of 1852 when the Marquette Iron Company shipped six barrels of ore
to New Castle, Pennsylvania. The following year, the Cleveland Iron
Company shipped an additional 152 tons of ore to Sharon, Pennsylvania
(Hickok et a1. 1938:17).

Those years from first discovery of iron ore to 1855 represent
an early experimental phase in the iron industry, when mining technology,
transportation, and production techniques were based on trial and error.
This was primarily because of the lack.of available information from

which economically profitable decisions could be made. At the center

35

’-

of this decision making process was the transportation problem, and this
single factor was of greatest importance throughout the duration of the
iron are industry (Hatcher 1950:62). For example, transportation
involved 1) transportation of ore from.the mines to either forge sites
or docks, 2) transportation of charcoal to the forge sites, 3) trans-
portation of finished iron (or ore) to southern markets, and 4) trans-
portation of are or iron from.aouthern docks to furnaces or mills.
These transactions include the problems of transshipment of each change
of transport. Before 1855, the iron companies experimented with their
various options--chiefly, whether it would be more economically
profitable to produce ore locally and ship finished iron to southern
markets, or concentrate on shipping bulk are only. The shipments of ore
in 1852 and 1853 were most likely trial runs testing out the latter
option.

As early as 1837, Governor Mason of Michigan wanted a canal
built around the rapids of St. Mary's River, but it was not until 15
years later that Congress finally authorized the construction of this
canal (Hickok 1938:17). Work on the canal began in June of 1853 and
was finally completed in May of 1855. Before the opening of the St.
Mary's Canal, only a dozen or so vessels sailed upon Lake Superior,
all of which had to be dragged over the portage around the Sault (Bald
1954:243). Many authors have discussed in great detail the actual
building of the canal at Sault Ste. Marie, and their thoughts, will not
be reiterated here in great detail (see Neu 1953:25-46). In general,
the completed canal consisted of two locks, each 350 feet long, 79
feet wide, and 13 feet deep (Bald 1954:244). The opening of the canal

at the Sault provided the first opportunity for any real economic

36
investment or development in the Lake Superior region. Before 1855,
vessels carrying industrial raw materials such as are or timber were
forced to unload their cargoes at the falls and transport them.across
the portage to awaiting vessels on the other side. Unloading and loading
in this manner was accomplished by means of wheelbarrows and hand labor,
and the costs were so expensive as to make any regular service of this
nature economically unprofitable.

The Sault Canal changed this economic picture entirely, totally
eliminating transshipment costs at the falls. Decision makers in the
iron industry then faced a new economic problem. Again, the question
was whether it was more feasible to produce iron at local furnaces or
concentrate upon shipping are to southern operations. The period 1855
through the Civil War was one of high investment in the iron industry
in mining, iron production, and transportation. For example, the mining
town of Negaunee consisted of six rough-hewn log houses and a dozen
workers in 1855 when the canal opened. Less than four years later, the
population had risen to more than 300 workers. After ten years of slow
growth, this population boom was a direct repercussion of the growing
need for iron and the opening of the Great Lakes Waterway through the
Sault (Hatcher 1950:81).

Population expansion in the mining areas of the Marquette Range
was a reflection of overall expansion of the mining and refining
industries. In 1855, a major push to inprove overland transportation
from the mines to Lake Superior (a distance of 10 to 15 miles) began. A
plank road from Ishpeming to Marquette was built that year and in 1857,
it was fitted with tracks for a steam railroad system owned by the Iron

Mountain.Railroad Company (Hatchet 1950:72). Improvements in

37

transportation of are greatly decreased freight rates from the mines to

Marquette and from.Marquette to the major Lake Erie ports. Table 2 shows

this decrease in rates from 1855-1858.

Table 2. Freight Rates (Hatcher 1950:82)

Year Mines-Marquette Marquette-Lake Erie Ports
1855 $3.00/ton $5.00/ton
1856 $1.27/ton $3.00/ton
1857 $1.27/ton $2.67/ton
1858 $ .87/ton $2.09/ton

This decrease in freight rates illustrates the impact of the Sault Canal
and improved local transportation on the economic picture at that time.

.The passenger trade also changed greatly with the opening of the
canal. The shipping industry soon regained the advantage held by the
railroads as far as the western traffic was concerned. This provided a
steady supply of immigrant labor to all parts of the Lake Superior shore-
line, so that this key element was never a major factor in limiting the
growth of the iron industry.

In 1856, the first movement of goods occurred from the Great
Lakes directly to Europe. This event is noteworthy in that America was
rapidly developing its international market at the time when earlier
industrial powers such as England were declining in world economic
influence. An implication of this is that the economic decision makers
in the Lake Superior region had-with the opening of the Sault Canal,
at least-the possibility of connecting with world markets at some
future point in time. Investment in.timber and mineral resources in-
creased across the region until the overextension of credit resulted in

the Panic of 1857, which caused a temporary but measurable slowdown in

38

growth. The impact of the "Panic" was hardest felt in the iron industry
by a marked depressed demand for iron and iron ore; even the Pittsburgh
mills were forced to temporarily close (Hatcher 1950:88). Fortunately
for the iron industry, this depression was short-lived and by the
following year, production was again on the rise. The panic did
illustrate the benefits of sound financial backing as well as the
sensitivity of the Lake Superior region to the nation.market system.

The beginning of the Civil War a few years later had several
impacts upon the early development of the iron industry in Michigan, the
most important of which was the overall stimulation of the market
because of increased demand for iron. Prices for are rose from $6.00/
ton in 1859 to $8.50/ton in 1864 (Mansfield 1899:566). Finished iron
prices in Pennsylvania were also undergoing marked increases, and at
Hopewell Village in southern Berks County, prices rose from $30.00/ton
before the war to $80.00/ton in 1864 (Walker 1966:64). These nationwide
increases helped stimulate production in the Lake Superior mines of the
Marquette Range, with are production increasing from about 69,000 to
280,000 tons over the seven years beginning in 1859 (Mansfield 1899:
566). These increases reflected the expansion of mining operations, as
well as improvements in the efficiency of ore movement.

Other influences of the Civil War on the iron industry were less
tangible but of equal importance to future development. For example,
there was a general improvement of opinion regarding Lake Superior iron
and iron ore. Before the war, the quality of iron from the range was
not widely known and therefore in no great demand. With the greater
exposure of the metal to the industrial sector this changed, and Lake

Superior region ores proved capable of producing high quality iron.

39

Another byproduct of Ihe Civil War was the greater exposure of the region
as a whole to the "outside" world, especially the usefulness of the
Great Lakes for transportation of goods. The war brought about
reevaluation when a blockade of the Mississippi forced a change in the
routing of goods from.the Midwest and central states through the Great
Lakes transportation system. The war generally familiarized the
commercial sector of the American economy with the true value of the
Great Lakes region in terms of transportation and mineral wealth
(Hatcher 1950:94).

Following the Civil War, the development of the Lake Superior
iron industry continued at a rapid pace. Even though the military
demand for iron fell, the direction of America toward industrialization
kept the need for the metal at a high level. The railroad industry in
particular lead in consumption of both charcoal and cast iron. While
eastern iron-producing states such as Pennsylvania lead in the pro-
duction of cast iron using anthracite coal and later coke (primarily
after 1875), the Upper Peninsula furnaces of Michigan produced only
charcoal iron (LaFayette 1977:VIII). This was because there were no
other real fuel alternatives within economic reach, since coal could not
be shipped north because of deterioration and moisture problems. On the
other hand, hardwood was locally available and most early iron companies
had highly diversified holdings and controlled large amounts of forest
land from which they could extract timber. After 1870, the requirements
of shaft mining forced the companies into the logging business in order
to supply support timbers for the underground caves. Subsequently,
charcoal iron.was the only economical alternative left to the companies

beyond the option of direct ore shipment.

40

The type of iron produced with a charcoal fuel was unique
because it contained a much lower carbon content than coal or coke iron,
making it less brittle. This quality was desirable for many early
industrial uses, one of which was railroad wheels. Later, the demand
for charcoal iron decreased to such an extent that by the turn of the
century, charcoal iron was of little importance.

During the period from.the Civil War to around 1880, a number of
major changes altered the course of the iron industry. Of greatest
overall importance was the shift in philosophy from.a mixed pattern of
iron production and are shipment to a full scale program of direct
southerly shipment of the raw ore. This change occurred gradually and
was accepted by different companies at different points in time. One
major reason for the changeover to are shipment was the increase in
expense of local iron production. Although 25 furnaces were built in
the Upper Penninsula during the second half of the nineteenth century,
almost all failed to make substantial profits as expected (Hatcher 1950:
96). Numerous factors accounted for these failures, but mainly the
general expense of charcoal iron. Since it took one acre of hardwood
timber (35 cords) to smelt 14 tons of iron, the cost of procurring this
resource rose with its increasing scarcity. By the turn of the century,
the few furnaces remaining in operation were consuming over 10,000 acres
of hardwoods each year (Hatcher 1950:102). As timber stands decreased
locally, transportation became a problem and charcoal-producing
communities, such as Rock Kilns in Alger County, Michigan, were
established along railroad lines to supply established furnaces
(Hulse et a1. 1977:4-7). Overtime, these increasing costs, combined

with lessening demand for charcoal iron, resulted in substantial losses

41

for many companies and little or no profit for others. But the idea of
local production was slow to die and before the last furnace shut down,
the Upper Peninsula had produced almost two million tons of iron.

Table 2, Appendix A provides a listing of Upper Peninsula forges and
furnaces, along with respective production figures.

Even though two million tons of finished iron was of major
economic importance to the region, this figure is small in comparison
to total regional shipments of bulk ore for the same period. From
1855 to 1930, five Lake Superior basin mining ranges produced over 150
billion tons of ore. Figure 3 illustrates the relative production of
the four ranges that transported ore across Lake Superior. Table 3,
Appendix A presents annual totals for these ranges.

As this graph illustrates, production rose steadily for all four
ranges as did the transportation of ore from the ports of Duluth,
Ashland, Marquette, and Two Harbors. The Duluth port in particular
showed drastic increases after 1895 and led all other ranges for the
remainder of the period.

The discovery and development of these five ranges is of great
importance in understanding the output and transportation of ore
throughout the Lakes. Before 1877, the Marquette range was the primary
production area for the region in terms of both finished iron.and iron
ore. Later discoveries of ore deposits in Wisconsin and Minnesota
changed the picture of the Great Lakes transportation as new railroads
were built and different ports developed to move the quantities of are
being mined. Table 3 gives a brief outline of the ranges, their dates

of discovery and development, as well as their major ports of shipment.

42

Figure 3. Iron ore production and transportation for the Lake Superior
Region 1855 - 1930

43

no

 

 

 

 

3 2 32 no 1 Ma me
939.8: 25. .‘AW ‘r‘
, <
«nuances: ( .
Suzi
:

 

 

 

 

mam.

o.

m—

ON

mm

on

mm

ow
no

we

 

8110:. nontm

44

Table 3. Lake Superior Iron Ranges

Range Discovered Opened Port

Marquette 1844 1854 Marquette
Menominee 1866 1877 Escanaba
Vermillion 1875, 1880 1884 Two Harbors
Gogebic 1881 1884 Escanaba, Ashland
Mesaba 1875 1892 Duluth

Sources: (Hatcher 1950:116-163; Bald 1954:278; Mansfield 1899:554-570)

Corresponding Figure 4 illustrates the location of these major iron
ranges and the flow of iron to southern ports.

As Table 2, Appendix.A demonstrates, there were relatively few
furnaces established on the new ranges. Those iron-producing areas
begun after the mid-1870s were located away from the ranges on major
transportation routes in areas of abundant hardwood forests or port
towns with easily accessible timber. This seems to indicate that by
1875, iron production.was regarded as a minimally important activity
compared to are shipment.

Expansion of mining operations at each of the five major ranges
had a gradual impact on the quality and quantity of ore that was removed
from these areas. All iron ore is not the same. Rather, in.many cases
it is geologically as distinctive as fingerprints on individuals
because of the uniqueness of the chemical composition of the ore bearing
strata. The composition of are changed as new areas were opened and as
the iron and steel industry developed new technologies for dealing with
various types of ore. Before 1870, the Marquette range shipped only
"hard" hemitite ores; the "soft" ores were not considered to have any
value until later (Mansfield 1899:555).

Iron ore as a mineral has a wide range of chemical compositions,

 

 

Vermi llion
Mesaba
Gogebic
Marquette
Menominee

U‘J-‘UONH

Figure 4. Iron range locations

45

46

 

 

 

3:2
'E

 

 

47

from.high grade ores with a large percentage of iron to lean ores with
large amounts of jasper, chart, and slate and less than 20 percent of
iron (Royce 1938:28). Varying amounts of iron carbonate, silica, iron
silicate, aluminum silicate, manganese oxide, magnesium carbonate, iron
sulfide, and lime phosphate also comprise most ores. The composition
of ore varies markedly within and between ranges and sometimes even
within a single mine. In general, there are two broad classes of ores,
namely "hard" and "soft." Within each of these classes there are five
subdivisions: 1) bessemer, 2) non-bessemer, 3) high phosphorus,
4) manganiferous, and 5) siliceous (Murray 1938:69). Hard ore is dense
and breaks into lumps whereas soft ore is earthy and more granular.
Bessemer ore classification depends upon the content of phosphorus in
the ore: to be regarded as bessemer, the ore can have no more than one
percent. Nonrbessemer ores therefore have more than this amount and are
less desirable OMurray 1938:69). High phosphorus ores contain from .18
to .25 percent phosphorus, and manganiferous and siliceous ores are
those containing high percentages of manganese and silica respectively.
In the early years of mining, the ore from a given.mine was
uniform in chemical composition,.but as.production increased less
desirable ore was shipped. By 1920, the mines shipped several different
grades of ore. Because of the need for regularity in the are at blast
furnaces, grading became necessary to assure certain specific composi-
tions of iron and trace elements in a shipment. Too much phosphorus or
too little iron would cause problems with the finished iron or steel.
The process of sampling ores at the mines, in the boat.cargoes, and at
the furnaces arose from this need for regularity (Murphy 1938:70-71).

Depending on the sample, the ore may be mixed with different

48

complementary grades to produce a desired composition. Before World War
I, grading consisted principally of separating bessemer from.non-
bessemer ores (Bayer 1938:73). Pricing was based on these grades and
only comparatively high-grade ores were-traded in the market place.
With the advent of specialized steels after World War I, the 10-part
grading process became more important (Bayer 1938:73-75). By 1940, the
ores removed from the Lake Superior region were so reduced in quality
that 40 percent were beneficated before shipment. Benefication is a
process that concentrates ore by crushing, screening, and washing to
improve the physical composition (Zapffe and Hunner 1938:77). By the
19608, this process resulted in the use of taconite pellets almost
exclusively as a form of ore shipment.

To summarize the types of ores that were shipped on Lake
Superior, Table 4 presents the chronological sequence of development of

graded and beneficated ores.

Table 4. Development of ore shipments

Date Ore Type

1845-1879 - Hard

1860-1870 Hard, Bessemer

1870-1920 Hard, Soft, Bessemer,
Non-Bessemer

1920-1940 Graded ores

1940-1960 Graded, Beneficated

1960-present Graded Taconite

The specific chemical composition of the ores, as well as this
general outline of ore shipment, can be used to assign a time period to
samples of ore from lost vessels. The availability of assay records is

essential to the linkage of are samples to specific mines, although some

'.

49

of this information is contained in publications of the Lake Superior
Iron Ore Association for various periods.

Transportation improvements in several sectors aided the gradual
shift away from local iron production in.Michigan's Upper Peninsula.
During the early years of experimentation in the iron industry, the only
transportation options available to southern ports were through the dock
at Marquette. Rail lines connecting the mines of the Marquette range
and the dock at the port rapidly boosted production. Likewise, later
railroad lines had a similar effect on production in areas that hereto-
for were inaccessible. For example, the other major ranges were not
opened until reached by major railroad routes to such ports as Escanaba
(Menominee Range; Chicago and Northwestern Railroad), Ashland (Gogebic;
Mfilwaukee, Lake Shore and Western), Two Harbors (Vermillion; Duluth and
Iron Range Railway Co.) and Duluth (Mesabi; Duluth, Missabe and Northern
to the Duluth and Winnipeg)(Hickok 1938:16-35). Iron companies and
interested investors financed these vital connections for the sole
purpose of moving the ore from the fields to the ports for southern
shipment. Other minor districts such as the Gunflint and Cuyuna ranges
shipped to docks at Port Arthur and Duluth respectively (Hickok 1938:
20, 23).

While railroad systems were developing in Michigan, Wisconsin,
and Minnesota, major improvements were also taking place in water trans-
portation. These improvements involved both the advancement of vessel
design and technology as well as the improvement of existing waterways
to accomodate increasingly larger ships. These changes further defined
the advantages of a strategy emphasizing shipment of ore over local pro-

duction of iron.

50

Nayiggtion and Industrial Expansion

From the opening of the Sault Canal in 1855 up to the present
day, changes in vessel technology have rapidly advanced in the direction
of increased size and capacity. In addition, vessel design has been
modified over the years to conform with modernization in areas such as
loading and unloading technology. The geography of the region has
always played a major role in determining the range of vessel types that
would be functionally useful to the shipment of ore to southern ports.
This is not to say that geography alone has been the single factor
affecting vessel design, only that it was a major (if not 525 major)
constraint to development in.most cases. For example, the Sault falls
constrained development of the Lake Superior region for many years until
the first canal system*was built. After completion of the canal, the
size of the locks affected vessel size in.much the same way-i.e.,
placing limitations on the size of vessels that could pass through the
locks. Table 5 lists the construction sequence of the canal network at

the Sault, along with the corresponding size limitations of each lock.

Table 5. Expansion of locks at Sault Ste. Marie

Year Opened Lock Name Maximum Length Maximum Width Maximmn Depth

1855 350 ft 70 ft 13 ft
1884 Weitzel 515 80/60 at gates 16
1895 Canadian 900 6O 17
1896 Poe 800 100 22
1914 Davis 1,300 80 23.3
1919 Pourrh 1,300 80 23.3
1943 MacArthur 800 80 30

Sources: (Bald 1954:245; Nute 1944:136; Hatchet 1950:112-114, 281)

The lock sizes presented in this table should be viewed not in absolute

terms because the vessels by necessity have to be considerably smaller

51

in order to pass the locks. Water depth.and vessel draft were major
concerns at both the locks and throughout the various passages from.Lake
Superior to Lake Erie ports.

Navigational improvements occurred at many points in time on the
Great Lakes. Regarding depth management, the increasing size of vessels
required channel dredging as early as 1855 in the St. Clair flats
(Manfield 1899:253). Before 1892, the maximum depth of connecting
channels was 16 feet-the same as the Weitzel lock at the Sault. From
1892 to 1897, a major channel improvement program was initiated that
created depths of 20 feet throughout the waterway from Superior to Erie.
Table 6 provides a brief listing of the major improvements and the dates
of completition. Depths before improvement can be viewed as environ-

mental constraints that placed limitations on vessel size.

Table 6. Navigational improvements

Date Area Prior Depth Depth after Improvement
1855-1858 St. Clair Flats 6 ft. 11 ft.
1871 St. Clair Flats 11 ft. 13 ft.
1874-1892 Detro1t River/ 13 ft. 20 ft.
Lime Kiln
1882-1895 Hay Lake Channel 6 ft. 20-21 ft.
1898 St. Clair Flats 13 ft. 18, 20 ft.

Source (Mansfield 1899:248-253)

When combined, Tables 6 and 9 show that the overall navigable channel
depths changed from six feet in 1855, to 11 feet in 1858, to 13 feet in
1871, to 16 feet in 1874, and finally to 20 feet in 1898. This implies
that these depths represent constraints on vessel size, and that the
accompanying dates provide a sequence of typological expansion. For

example, before 1871 the maximum.draft of a vessel traveling from Lake

52

Superior to Lake Erie was under 11 feet. This placed real constraints
on the design of vessels and the accompanying tonnage those ships could
carry.

A general increase in vessel size for ships engaged in the move-
ment of bulk cargo closely paralleled the sequence of channel depth
improvements. The development of vessels for the iron trade has been an
ongoing process since the first mines opened in the Upper Peninsula.
Generally, before the opening of the Sault Canal in 1855, the small
amounts of pig iron and iron ore that were shipped went by schooner,
although the screw propellor Vandalia was on Lake Superior by 1841
(Nute 1944:133). Although records for this early period are scarce, few
if any vessels engaged in the iron trade were lost during this time.
With the opening of the Sault Canal, schooner traffic increased
dramatically and was the principal vessel type engaged in the iron
trade until the mid-18603. Beginning in the 18603, steam barges
Operated on Lake Superior, making the first inroads in heavy coarse
freight. These barges Operated in a series of tows by which sail-
assisted barges were towed behind the steam tug barge. Older and less
efficient sailing craft were soon converted into barges and used in the
movement of coarse freight such as iron ore and lumber. This system of
tows that proved that the larger the cargo, the less expensive the
freight rate. This piece of information influenced the entire course of
shipbuilding in succeeding years (Nute 1944:125-126).

In 1869, the first true steam barge, the R.J. Hackett, was built

 

to serve the iron are trade. This vessel was constructed specifically
for carrying ore, with a design that emphasized the bridge in the front

and the engines in the rear (Barry 1973:107-109). This ship was a

53

combination of steam and sail, with the latter serving as an auxiliary
power source for open sailing when the winds were favorable. The
central portion of the vessel was open for cargo storage, with large
hatches spaced 24 feet apart, providing access to the hold for use in
the ore pockets at the newly built marquette docks. Numerous companies
built woodenrhulled vessels similar to the R.J. Hackett until the late
18808 when ironrhulled ships were first produced.

Ship builders began producing ironrhulled vessels after the
successful service of the 92252, built in 1882 as a prototype vessel
(Bald 1954:279). The 93252 was the first of the ironrhulled bulk
freighters and the largest on the lakes at 282 feet in length and a
carrying capacity of 2,164 tons. The iron hulls allowed for larger,
more stable ships and soon after the Qggkg was launched, iron became the
material used in hull construction. Because of the peculiarities of the
insurance system.at the time, the first iron vessels were of composite
construction and were iron only from the deck to the water line, with
wood below. With the use of iron, bulk freighters were built in numerous
styles and evolved gradually toward larger sizes within the constraints
imposed by navigation at the locks and channels between Lake Superior
and Lake Erie ports. As steel freighters began to be built in 1886, the
average size of bulk freighters again increased to just over 300 feet in
length, with capacities of 3,400 tons.

By 1906, 600-foot freighters had become the standard workhorses
in the Great Lakes iron ore trade (Barry 1973:173). There were several
reasons for the replacement of sailing ships with increasingly large bulk
freighters. The number of sailing ships, including schooners, barks,

and brigs, declined steadily from 1870 to 1890. The main reasons for

54

this were: 1) size limitations of vessel design, 2) lack of maneuvera-
bility in narrow channels, 3) lack of reliability in length of trips
from.Lake Superior to Lake Erie, and 4) increased cost of wages for
sailors working on sailing ships (Barry 1973:143). These four factors
made sailing ships uneconomical to operate in comparison with the newer
and larger iron vessels in the ore trade.

From 1893 to 1897, a major economic depression bankrupt many of
the small iron companies operating in the Great Lakes. The companies
that remained solvent through the crisis gained a tremendous competitive
advantage and eventually agglomerated into three major corporations-
Consolidated Iron Mines, owned by J.D. Rockefeller; Minnesota Iron
Company (which became U.S. Steel in 1901), backed by J.P. Morgan; and
Oliver Iron Mining Company, owned by Andrew Carnegie and others (Barry
1973:174-177). This consolidation formed large corporations with tre-
mendous financial capabilities and an equally large desire for as large
a piece of the market as possible. Economics dictated that it was
financially most advantageous for these companies to own their own ore
shipping fleets rather than to pay an outside shipper. Eventually this
situation resulted in a major period of shipbuilding that lasted through
World War I to the next economic downturn in the mid-19203. During this
period, several types of vessels were put into use in the ore trade,
among which were the whaleback freighters and the "standard" 600-foot
line of bulk carriers. The former were comonplace from the late 1880s
to their eventual elimination by 1940. These whalebacks were iron
hulled, flat nosed, and similar in appearance to modern submarines.

They were eventually supplanted by the larger bulk freighters that could
accomodate changes in loading technology and hence more economical to

operate (Nute 1944:125).

55

Table 7 provides a brief summary of vessel types for those ships

used in the iron trade.

Table 7. Development of Great Lakes bulk carriers

Nominal Capacity
Date Type Length-Ft. Long Tons
1855-75 Wooden sailing ships, 200-250 300-700
schooners
1860-75 Steam barges, sailing tows
1870-85 Wooden hulled freighters 210 1,200
1882 Ironrhulled steamer 300 3,000
1888-1930 Whalebacks
1895 Steel-hulled steamer 400 5,800
1900 Steel-hulled steamer 500 8,200
1907 Steel-hulled steamer 600 12,000
1938 Turbine-powered coal 610 14,000

Source (Erickson 1969:203)

Although this sequence of vessel types is quite simplistic, it shows
the general changes through time. Journals such as the Marine Engineering
Record should be consulted for the fine details of freighter construction
and evolution.

The evolution of larger bulk carriers caused a major reduction
in iron ore freight rates from the Lake Superior and Lake Michigan docks
to Lake Erie ports. Figure 5 graphically illustrates this rate
reduction from 1854 to 1920. The increased capacity of ore-carrying
vessels brought about this rate decline. The major effect of this
gradual downward shift in rates was to eventually convince the decision
makers controlling the iron industry to shift efforts to shipment of ore
rather than local production of iron. As Figure 4 illustrates, a major
reduction occurred beginning in 1874, the same date as the last major
period of furnace building in Michigan's Upper Peninsula (see LaFayette

1977:49).

56

Figure 5. Freight rates for iron ore 1855-1928, Marquette to Lower Lake
Ports (From Hickok 1938:324)

57

mm mm

m—

mom—

mm

use»

mm

ms

no

unm—

 

 

 

 

 

om

oc—

on.

com

omN

com

own

ace

aOU\w

 

58

As the industrial economy of the Great Lakes region expanded,
ship traffic on the lakes increased at phenomenal rates. In order to
control the flow of traffic as well as to promote safe navigation on the
lakes, the federal government was constantly involved in the development
of the Great Lakes transportation system. The Lighthouse Service,
established in 1789, was one governmental organization dedicated to the
improvement of navigation on all the nation's coastlines, including the
Great Lakes. This service became involved on the lakes in approximately
1820 when the first lights were established at Presque Isle and
Buffalo (O'Brien 1976:13). The earliest lights were built on the lower
lakes and as traffic expanded into the upper lakes, the lights soon
followed. Lake Superior, for example, was without lights until 1847-
only a few years after the iron and copper industries began mining
operations in the area. The lights established across the lakes were
placed at strategic points to outline hazards or to mark common
navigational points such as harbor and river mouths or points of course
change. For further information on the history of specific lights, see
O'Brien (1976) and Holland (1972).

In 1876, the Lifesaving Service also began operating on the
Great Lakes. The Great Lakes became the nineth, tenth and eleventh
districts within the national system, and headquarters were established
at numerous points along the Lakes to provide a rescue team for ships
in distress. This service was responsible for thousands of rescues, as
was the organization it later became--the U.S. Coast Guard.

A.major area of governmental concern during the development of
the Lake tranSportation system.was directed toward the creation and

improvement of harbors. Numerous congressional appropriations were

59

were approved for this task, with the goal of reducing vessel loss
through the availability of refuge harbors (Mansfield 1899:263-264).
Table 8 provides a list of harbors on the upper lakes as well as dates
of improvement, if any. This listing covers only those improvements

before 1900 (Mansfield 1899:315-364).

Table 8. Harbors of Lake Superior

Port Date Navigable Depth-Feet

Sault Ste. Marie
White Fish Point

Grand Marais, MI 1881-1898 13
Munising

Marquette 1867-1898 18-40
L'Anse

Copper Harbor

Eagle Harbor 1879 14
Ontanagon 1867-1898 12
Ashland 1886-1898 16
La Pointe

Bayfield

Portwing

Duluth l87l-1898 16
Superior 1867—1898 16
Two Harbors 1889-1898 good navigable depth
Grand Marais, MN 1880-1898

Just as the depth of navigable waterways was of great importance
to the development of lake vessels, so also was the size and depth of
harbors--especially those used as primary loading and unloading docks.
The dates of harbor improvements closely parallel the development of the
various regional industries, and many harbors had ongoing programs of
expansion in order to maintain facilities for increasingly larger
vessels CMansfield 1899:354-362). Harbor improvements included the
dredging of harbors, the building of breakwaters and dock systems, and
the cutting of channels into nearby smaller lakes. Although many

harbors underwent some type of improvement, most were not suitable for

60

use as harbors of refuge because of their unprotectedjnature. Only
those harbors that could provide safety in storms were recognized as
refuges, and these were extremely scarce on the lakes before massive
breakwaters became commonplace. Additionally, the gradual expansion of
vessel size made only certain harbors available for use either as
normal harbors or refuges. Depth and size of harbors were again the
major constraining features for large vessels used in bulk ore shipment.
Therefore, a large vessel in distress was severely limited as to which
harbor could be sought without fear of running aground.

At the major ports handling the shipment of iron ore, numerous
changes took place in both harbor improvements and technological innova-
tions in dock technology. Techniques of unloading, for example, kept
pace with the evolution of lake vessels, whereas loading and unloading
took place with ore in barrels and then later in bulk by shovels and
wheelbarrows. The invention of the clamshell bucket just before the
Civil War greatly reduced the time and costs of ore transfer (Hatcher
1950:114-115). Later improvements involved the construction of rail-
road trestles and docks so that vessels could be loaded directly from
‘mine cars. These improvements have resulted in a gradual reduction in
the length of time needed for the loading and unloading process. At
the turn of the century, the average turnaround time per 1000 tons of
cargo was 331 minutes. By 1950, this time had been reduced to 37
minutes (Hatcher 1950:217). In terms of average total loading time, the
length of stay at the dock decreased from 52.5 hours in 1900 to 5.6
hours in 1950 (Hatchet 1950:217).

The overall effect of technological improvements in vessel

design and dock facilities was to increase the number of round trips

61

that vessels could make in a season. In addition, for approximately the
same period, there was a phenomenal rise in total tonnage on the lakes
but only a slight or moderate increase in the average number of working
vessels (wolff 1972:62-63). The combination of technological improve-
ments in docking and in increased size led to an overall gradual
reductiOn in the number of vessels working the lakes. This was
especially true in the iron industry with the evolution of bulk ore

carriers after the Civil War.

The Grain Trade

The Great Lakes grain trade is comprised of numerous agricultural
commodities that are often treated as a single commodity-grain. But in
reality, raw products like wheat, corn, oats, barley, rye, and processed
cereals and flours are each separate entities with special considera-
tions. They are grown and processed in different regions at different
points in time and have particular qualities that affect their preser-
vation and transportation, making each commodity unique. Traffic in
these different grains centers on various markets across the Midwest,
with each marketplace dealing in varying amounts of certain grains.

In the Lake Superior region and surrounding hinterlands, wheat
has historically been the dominant commodity, with near exclusion of the
other types of grains. Since the 18708, wheat has comprised a large
majority of the total grain traffic on Lake Superior (Williamson 1977).
This dominance of wheat is a result of the environmental conditions
present in the region as well as the frontier atmosphere that permeated
the region into the 19308.

During the period between the opening of the St. Mary's Ship

Canal and the early 18708, the Lake Superior region underwent a process

62

of settlement and development in all sectors of the economy. The emerg-
ing industries such as iron, copper, and lumber rapidly settled the
region. This early period also saw experimentation in agriculture in
Michigan, Wisconsin, and Minnesota, with special emphasis on wheat pro-
duction. The flow of grains before 1870 moved in an upbound route with
wheat and flour coming from southern and eastern areas to meet the food
demands of the developing region (Mansfield 1899:526). Ships would stop
at each of the accessible ports with cargoes of flour and whole grains
(for animal feed) and then return to the south with varying loads of
freight.

The early 18708 mark a turning point in the grain trade and the
emergence of a new pattern of commerce on the upper lakes and Lake.
Superior region (Odle 1953d:260). Before this time, the center of the
national grain trade was in the north central region of the country,
with the Great Lakes forming the north and east boundaries and the
Mississippi and Ohio rivers forming the south and west borders
respectively. The eastern transportation route to the New England and
Middle Atlantic states opened upon the completion of the Erie Canal in
1825, and grain found it's way from.Midwestern farming centers along this
route. Later, canal networks such as the Ohio and Erie, Miami and Erie,
Habash and Erie, and Illinois and Michigan opened up this eastern route
to interior agricultural areas (Odle 1951:238-239). Cleveland, Toledo,
Detroit, Chicago, and Milwaukee became the major market centers for this
early trade.

At this same time, Midwest grain products were also moving south
by way of the Mississippi River system. Like the east, the southern

portion of the country before the Civil war could not feed itself

63

without importing food stuffs from other regions. The flow of grain
from.the southern portion of the east North Central region was accom-
plished through St. Louis and Cincinnati markets. From there, grain
moved south to New Orleans and on to Southern Atlantic states (Odle
1951:245). In the 18508, some of this Mississippi River trade was
diverted through the eastern Great Lakes route because of the settlement
of northern Great Lakes states and the commercial revolution brought
about by the development of the railroad and telegraph (Odle 1951:244).

By 1850, a westward shift in the grain trade was underway and
Toledo, Chicago, and Milwaukee became the focus points of the market as
Cleveland had been in earlier years (Odle l952a:24). The decades of the
508, 608, and 708 were especially important to the grain trade because
during these years, the centers of flour and grain marketing underwent
rapid change largely because of improved transportation and communication
between the West and East. The telegraph, for example, facilitated
direct purchases of western commodities by eastern buyers. Transporta-
tion by newly deve10ped water and railway systems allowed for the
reliable movement of grain commodities to eastern centers, thus by—
passing the middlemen. Before the 18508, most grain was funnelled into
Buffalo and from there, distributed to the East. The railroad and tele-
graph allowed Buffalo to be circumvented, the individual eastern market-
places could directly purchase western grains (Odle 1953b:108).
Ultimately, this resulted in a westward shift of grain and flour
markets.

During the Civil war, railroads successfully competed with.
water transportation for the first time. Normally, railroad movement

of grain was reserved for winter transportation when the water routes

64

were closed for the season. The need for foodstuffs for the Union war
effort necessitated the rapid deployment of grain and flour by rail
routes. After the war, the "normal" system of grain movement was
resumed and water transportation remained immensely more economical than
rail transport for most agricultural commodities. This was not the same
for the movement of flour, which continued to be transported by rail
even after the war ended. Flour was the first commodity of comparatively
low value the railroad carried that could successfully compete.with
water transportation. Several factors accounted for this success, in-
cluding the difficulty of transshipment and the need for rapid movement
of this perishable commodity. In addition, flour could not be stored
over the winter months without considerable loss, so railways were the
only alternative to transportation for flour made from the fall harvest
(Mansfield 1899:530).

In the Lake Superior region, the grain market was not well-
developed before the early 18708. The movement of markets in a westward
direction, combined with western land settlement, eventually led to the
opening of the Lake Superior region to traffic in wheat and other grains.
The wheat that traveled on Lake Superior came from two major sources-
one American and the other Canadian. The American grain trade on the
lake began with the production of wheat in northern Minnesota and
eventually moved to encompass portions of the Dakotas (Odle l953dz260).
Wheat produced to the south of this area found its way through Lake
IMichigan ports, such as Milwaukee and Chicago, and did not have to
travel across Lake Superior.

The first wheat produced in Minnesota in the 18508 was shipped

south to St. Louis by the Mississippi route and east to New York

65

through ports such as Milwaukee (JarchOW'l948:4—5). This pattern
remained the same until the Civil War, when the Milwaukee port gained
major control of the market. Despite competition from Chicago,
Milwaukee retained the major share of Minnesota wheat traffic until the
'18808.

In September 1870, the grain trade on Lake Superior began with
a railroad connection from St. Paul to Duluth, thus opening the traffic
of the inland areas to the northern water route across the lake. In
the fall of 1871, the influence of Duluth as a shipping port was
further enhanced when the St. Paul and Pacific Railroad reached the
town of Breckenridge in the upper Red River Valley, and the Northern
Pacific Railroad secured a connection between Duluth and Moorehead
(Jarchow 1948:13). This latter route was completed to Portland,
Oregon, in 1883 (Garraty 1971:75). These railroads tremendously
increased settlement and agriculture in the routes, all of which were
channeled through Duluth. In the 10 year period from 1870 to 1880,
wheat production doubled with much of this increase coming from the
newly opened lands (Jarchow 1948:16).

Wheat production was not limited to Minnesota, but began a
general westward movement along transportation lines. Wheat adapted
well as a frontier crop because of its relatively low operating expense
per return. With.large.amount8 of inexpensive land available in
frontier areas, the problems of declining fertility were of no great
consequence. But as the land became more settled and prices of farm
land rose, wheat became less and less profitable in relation to other
crops or land uses. For this reason, wheat has historically followed

the westward settlement as a frontier phenomena (Jarchow 1948:27). Thus,

66

the movement of wheat areas first from southern Minnesota to the north
and eventually into the Dakotas and westward can be explained in part

by this concept of changing rent values. Nonetheless, the importance

of expanding wheat-growing areas to the Lake Superior grain trade
remains because Duluth became the single terminus point on the lake from
which these commodities were shipped.

The American grain trade on Lake Superior was not completely
restricted to wheat, although this was by far the major grain. Wheat
flour can also be regarded as a major part of this overall commodity.
Flour milling, like wheat production, also had its major beginning in
the state of Minnesota. .Minneapolis became the center for this pro-
duction and throughout the period, lead in flour production for the
Lake Superior trade.

Before 1873, the flour produced in the region was of relatively
low quality and brought prices commensurate with its position. The
major problem with.the flour quality lay in both the varieties of wheat
being processed and also the technology used in flour production.
Because the climate of northern.Minnesota and the Dakotas were too
severe for the existing winter wheat varieties, spring wheat was the
primary type grown in the region. The hard, red spring wheats were
well-adapted to the cold climate, but the bran of this type of wheat
shattered during the milling process and imparted an undesireahle orange
tint to the flour (Olde l953d:260). The technology of flour production
consisted of grinding the wheat between rapidly spinning millstones so
that as much flour as possible could be.made at the first grinding.
When this process was used, it did not fully succeed in separating the

various components of the wheat berry. And as a result, the bran

67

discolored the flour and the flour sometimes became rancid from excess
oil in the germ. In addition, the process often sifted the gluten cells
out of the flour, which adversely affected its baking quality.

When spring wheat was milled, these problems became more
exaggerated and a very low quality flour was produced (MacGibbon 1932:
402). Winter wheat flour was whiter, stronger, and had better keeping
qualities than spring wheat flour, and therefore it commanded a
premium price.

In the early 18708, Minneapolis millers imported a European
technique of refining that eliminated the specks of bran while pre-
serving the high gluten content of the wheat. This technique, called
the middlings purifier, allowed for the deporation of the bran before
grinding. A system of slow grinding first cracked the wheat. Then the
remaining berry (middling) was purified from.the bran and ground into a
pure white flour (MacGibbon 1932:403). The resulting new "patent" flour
established a large market for hard spring wheat. Wheat processed with
the middlings purifier was immediately more desirable, and prices
tripled immediately (Jachow 1948:16; Odle l953d:260-26l). This in turn
stimulated more spring wheat production, and encouraged further expansion
of the frontier region.

Other technological innovations of major import to the milling
industry included the adoption of the metallic roller process of
grinding and the centrifugal reel for sifting (MacGibbon 1932:403).
These innovations continued to enhance the quality of flour produced
and strengthened the financial base of both the American and Canadian

flour and wheat industries.

68

With all the wheat and flour of the north central region passing
through the ports of Duluth and Superior, large grain-handling facilities
quickly developed. From the 18708 on, grain elevators, storage ware-
houses, and loading technology expanded in tune with production.

Because of the seasonal nature of the wheat crop, storage facilities
were an essential element of the grain trade, and by the turn of the
century, there were 18 major elevators at Duluth/Superior with a
capacity of 25 million bushels (Mansfield 1899:536). Wheat began to
ripen in early and mid-August and continued into Navember, depending on
local conditions (Jachow 1948:21). Grain and new flour traffic on Lake
Superior was heaviest during these last months of navigation before
winter. Since flour could not readily be stored over the winter that
which did not leave port in the fall was shipped by rail during the
winter months. Upon spring re-opening of the Lakes, wheat was one of
the first commodities moved and in the months of April and May, grain
traffic was quite heavy while flour traffic was slight to nonexistant.

In contrast to grain, which was transported by bulk carrier,
wheat flour was usually shipped in barrels, and therefore often trans-
ported by package freighters or vessels carrying mixed cargoes.

Package freighters often made extra stops along a route to pick up or
drop off cargo, and the trade routes of these vessels were different
from bulk carriers, with only a single destination.

The second major component of the Lake Superior grain trade was
the Canadian system of production and transportation. This Canadian
grain industry evolved along separate but parallel lines and was closely
linked throughout with changes in American markets and policies. In

Canada as well as the U.S., the pattern of wheat agriculture that

69

followed frontier expansion was the same (MacGibbon 1932:3-16). Farming
in Ontario and eastern provinces began during the French Regime but did
not reach high rates of production until the nineteenth century.
Stimulated by increased demands for exports during the Crimean War
(1853-1856) and the American Civil War, wheat farming in Ontario and
Lower Canada rose sharply (MacGibbon 1932:17).

During the pre-l847 period of grain production, Canada shipped
nearly all its grain through.Montreal for export to Europe. It exported
very little grain, if any, into the U.S. either for consumption or trans—
portation. After 1847, political agreements between Canada and the U.S.
resulted in the opening of transportation networks previously restricted
to sole use by each respective nation. A rapid change in Canadian grain
shipments accompanied this agreement, and large amounts of wheat and
other products were exported from Upper Canada through American seaports
on the Atlantic (MacGibbon 1932:18).

As Canada became more densely settled, the wheat farming areas
began to shift to the west into the provinces of Manitoba and
Saskatchewan. Although farming was attempted in these areas throughout
the nineteenth.century, it was not until the 18708 that wheat production
became of any great importance. This was because of the lack of
efficient transportation systems to move grain eastward. After Duluth
secured its railroad connections in 1870 and 1871, grain from.Canada
moved by inland waterways to junctions on the American route to the
East. From Duluth, Canadian wheat made its way across Lake Superior,
through the St. Mary's Canal to Canadian ports on the Georgian Bay, and
then eastward by rail. Preceding 1883, grain from Manitoba was

channeled through Duluth and across the same routes as Minnesota and

70

Dakota grain (MacGibbon 1932:26-28).

A.major grain transportation revolution occurred with the come
pletion of the Canadian Pacific Railroad in 1883. This railway connect-
ed Winnipeg and Port Arthur, giving access to a Canadian port on Lake
Superior. Although hindered by financial difficulties in early years,
the Canadian Pacific was eventually completed to the west coast in 1885
(Royal Grain Commission 1938:18). This railway opened fertile western
lands and wheat agriculture closely followed settlement in these western
regions. Wheat production steadily increased in the Canadian west as
settlement and transportation became widespread. The Great Northern
(1889), NOrthern Pacific, and Grand Trunk Pacific railroads supplied the
means for the movement of immigrants, goods, and pioneer wheat from
these western areas (Royal Grain Commission 1938:16-20). During 1883
to 1920, the twin cities of Port Arthur and Fort William constituted
the major grain shipping centers for the entire western Canadian grain
trade.

The opening of the Panama Canal in 1914 caused far-reaching
changes in the flow of ocean traffic, including the world movement of
grain. Farmers in Alberta and western Saskatchewan now had an alter—
native route for their exports CMacGibbon 1932:266). Despite a delay
caused by World war I, by 1920 there was a diversion of western grain
away from the eastern Great Lakes route to a Pacific route leading
through.Vancouver, British Columbia. Lack of grain-handling technology
at Vancouver and initial resistance from.the railroads made growth slow
but steady throughout the 19208.

The Canadian milling industry grew along lines parallel with

American flour production. Technological innovations introduced into

71

Minneapolis soon found their way into Canada and had similar successes.
In Canada, milling became a major activity in several towns along the
route from the western prairies to the eastern seaboard. Along the
Canadian Pacific Railroad, the towns of Winnipeg and Keewatin (130
miles to the east) became major milling centers for western grain
(MacGibbon 1932:404). Flour would then be shipped eastward to Duluth
and across Lake Superior or would travel an all-rail route through one
of numerous connections east.

Other milling centers include Montreal and Port Colbourne, the
latter serving as the entrance port of the welland Canal at the eastern
end of Lake Erie. Lesser areas of flour production also include the
towns of Goderich and Fort William (MacGibbon 1932:405). Transportation
of flour produced at Fort William was also shipped on Lake Superior

following a southerly course to Sault Ste. Marie.

The Coal Trade

 

The coal trade on the upper Great Lakes differed greatly from
the other commodities moved in this region primarily because coal was
an upbound cargo moving from Lake Erie ports northward rather than
southward, as did the other cargo types. Another major difference lies
in the distribution of coal along this northerly route. Unlike ore,
grain, and lumber, which were produced in an area and shipped to
another specific area, coal was produced and shipped to a much.wider
variety of destinations. Rather than a raw material that was processed
in one central location, coal was a fuel source that needed no further
processing. As the need for fuel grew with expanding population and

industry, greater amounts of coal were shipped into the northern lakes.

72

And as new settlements grew and exhausted the timber resources of the
surrounding countryside, coal became more economical and trade routes
to these towns sprang up.

Coal had many uses for the Lake Superior region as its population
expanded. From their earliest beginnings, for example, the mining
industries were highly dependent upon steam power. The transportation
industries of shipping and railroads also needed coal for steam power,
and with the railroad boom in the late 18808 and 18908, demand for coal
sharply increased. The railroad industry was especially dependent upon
coal since steam.powered vessels could not afford to carry bulky wood
fuel on board. A steamship traveling from.Lake Erie to Duluth.had to
carry enough coal for both upbound and downbound legs of the journey.
So much of the coal that passed through the Sault Canal never touched
land because it was consumed by the vessel throughout the course of
travel (Williamson 1977:186).

Other coal destined for population centers was deposited at the
major ports along Lake Superior (Williamson 1977:179). In fact,
because of the region's settlement pattern, these lake ports remained
the only major population centers for the entire 1855 to 1930 period.

Those major ports and their dates of settlement are listed in Table 9.

Table 9. Lake Superior ports

Name of Port Settled Major Development Period
Grand Marais, MI 18603 1860 ~ 1910
Munising, MI 1867 1867 ~ 1920
Marquette, MI 1845 1857 - present
Houghton,‘MI 1852 1873 - 1930
Ontonagon, MI 1852 1855 - 1900

Ashland, WI 18508 1877 — present
Duluth/Superior, MN 18503 1871 — present

Two Harbors, MN 1884 1884 - present

Port Arthur/Ft. William, ONT 1884-92 1884 — present

73

iAlthough numerous other small ports were scattered along the Lake
Superior coastline, they most likely received few if any shipments of
coal.

Because of the numerous destinations of coal-carrying vessels,
the trade routes these vessels traveled were quite different from those
of the other bulk cargo carriers. The vessels carrying coal often made
several stops before emptying the cargo hold of a vessel. For example,
a vessel might make stops at Grand Marais and Munising before finally
unloading at Marquette. This diversity in destinations meant that the
vessels carrying coal were also more diversified than grain and/or ore
vessels. Smaller vessels were needed to enter the shallower ports, and
some coal destined for smaller ports was carried on vessels that
'accommodated mixed or package freight. Therefore, coal was not handled
‘specifically by bulk carriers that left those ports with ore or grain.
‘Upbound, these big carriers more often than not carried coal cargoes
-to increase the economic utility of the vessels. The large majority
(probably 80 percent) of coal brought into Lake Superior was destined
for the major ports of Marquette, Duluth, Ashland, and Houghton.
Initially, the coal was used primarily at those locations, but as the
railroads connected ports to interior areas, increased amounts of coal
were shipped out from these ports to the outlying areas. Thus, there
was a gradual change in the role of these towns from consuming points
to both consuming and distribution points. This was especially true for
Duluth, which acted as the center for almost all coal destined for the
northeast (Mansfield 1899:551). Table 14 lists the amounts of coal
that came into Lake Superior for the period 1855 to 1920. This table

shows that the major shipments of coal to the region began during the

74

late 18708. This corresponds with the time of major development in the

region, most notably the railroad era.

Historical Summary and Comparisons

The iron, grain, and coal distribution networks on Lake Superior
were similar in some respects and different in others. The major
similarities between these three commodities lie with the bulk nature
of the cargoes. Iron ore, grains of all types (excluding flour), and
coal were all loaded in bulk into vessels specifically designed to
carry these types of commodities. The cost of transportation for all
three commodities was dependent upon the amount of cargo moved in a
single voyage. The general types and sizes of vessels employed in
these trades would therefore be more similar to each other than they
would be to vessels moving other types of commodities. But this is
where the similarity ends and major factors distinguish the vessels of
each commodity trade from one another.

Trade routes are a function of three factors: 1) point of
origin, 2) point of destination, and 3) geographic space across which
movement takes place. These factors were significantly different for
each industry. The location of iron deposits and their chronological
differences in discovery and development contributed to a pattern of ore
shipment distinctive to the iron industry. The ports of Marquette
(1855), Ashland (1884), Two Harbors (1884), and Duluth (1892) became
the points of origin for tremendous quantities of iron ore. The
routes between those ports and Sault Ste. Marie were different from
grain or coal routes. Grain, for example, was not shipped on Lake
Superior before 1870 and only two major routes, Duluth (1871) and Port

Arthur (1884), were utilized for grain shipments. Compared to iron,

75

these chronological and spatial differences were quite significant, the
one excpetion being shipments from Duluth for the period 1892 to 1920,
when both used identical routes.

The coal trade is poorly documented in comparison to either
grain or iron industries, so comparisons must be made with extreme
caution. Since coal was often used as an upbound return cargo, the
routes traveled would be similar to a combination of the routes of both
the iron and grain trades, but would be dissimilar to either when
viewed separately. In addition, the destinations of some coal vessels
would not necessarily be to the same port or ports from which the down-
bound voyage was made. Smaller ports having dock facilities that could
accommodate bulk carriers would act as secondary distribution points
for the hinterland. Therefore, routes of coal delivery could require a
larger number of stops per trip and would entail deliveries at a larger
variety of port towns. Resulting patterns of coal trade distribution
systems would be significantly different from either iron or grain
patterns. The development of coal traffic on Lake Superior was closely
linked to the economic and demographic development of the region. The
trade routes that coal followed would therefore parallel the movement of
industry and the settlement of large populations, as discussed previously
in this chapter.

Along with differences in trade routes, a host of factors had
an effect on shipping, including: 1) the relative distance of each
route; 2) the presence or absence of navigational hazards, constricted
passages, or Course change areas; and 3) the weather patterns that may
affect one part of the lake differently from another. Each of these

three elements combine to intensify differences between routes. When

76

it is pointed out that iron ore routes are different from.grain routes,
the implication is that each industry will be faced with different
selective pressures.

Differences between iron, grain, and coal extend beyond routes
of trade and chronological development and into more specific types of
characteristics. The relative frequency of traffic, for example, is an
important factor in distinguishing between the three commodity trades.
Since vessel types and sizes were comparable among the industries under
consideration, the frequency of traffic of each.commodity would be a
function of the relative amounts of that cargo being moved on the lake,
as shown in Figure 6. Assuming that ore vessels and grain vessels are
the same size and type, then if one million tons of iron ore was moved
in a given period compared to a half million tons of grain for that
same period, the frequency of traffic for ore would be double that of
grain. That is, twice as many trips would have been made by ore vessels
than by grain carriers. Since this is potentially important to under-
standing vessel loss, the production figures previously listed in this
chapter assume an important role in determining the relative
frequencies of traffic between the three types of commodities.

When reviewing differences between the three commodities, it is
also necessary to examine the nature of the base industries that
controlled the production and shipment of these cargoes. For example,
the iron industry was highly centralized and developed with controls
over both.mineral extraction and refining systems. It was a type of
industry that had tremendous capital at its disposal and could therefore
afford the newest technology available. As vessels became larger and

more cost efficient, the iron industry invested heavily in these newer

77

Figure 6. Iron, grain, and coal tonnage on Lake Superior 1875—1910

78

o—

no

coa—

mm

was»

ca

 

 

 

mm

om

mum—

 

 

o—

m—

om

mm

on

mm

ca

no

81101 nontm

79

vessels. So it is likely that orewcarrying vessels were more modern
(i.e., larger) than other less centralized industries. In comparison,
the grain trade was more a comglomeration of smaller, agriculturally
based distributors less lflkely to act as a collective whole where
capital improvements were concerned. Vessels of the grain trade were
independently owned and operated as opposed to iron industry vessels
that were predominately (especially in the later periods) company
owned. The same could be said for those vessels involved in the

coal trade, except those in iron company fleets. The disparity in
capital inputs between the iron industry and other commodity trades is
great and would be reflected in the types and sizes of vessels in the
respective industries.

The nature of the cargoes is another factor that results in
differences between vessels of each industry. Iron ore, coal, and
grain have different qualities that affect how each is handled and
transported. Iron ore is impervious to water, whereas coal absorbs
water, which reduces quality and utility. On the other hand, grain is
often totally destroyed by water or is at the least rendered unfit for
human consumption.. The relative densities and bulk of these goods may
also have an effect on the likelihood of cargoes shifting in the holds
of vessels althougb.the specifics of this are not currently known.
Thus, the characteristics of the cargoes themselves may have been an
important factor in their shipping or salvage.

The seasonality of the various cargoes would also have created
differences between the movement and distribution of the three commod-
ities. Grain, for example, matures from mid-August to November and was

predominantly a fall crop. The need to move this grain before the close

'80
of navigation caused a surge of traffic of this cargo late in the season.
This situation was somewhat alleviated by the construction of elevators
that could store grain over winter months. Yet the traffic remained
heavy in the fall months, while a second surge was occasioned by the
opening of spring navigation and the need to quickly clear elevators
before spoilage occurred. This seasonality contrasts with the iron and
coal industries, which maintained steady shipments throughout the
season. Coal traffic may have been slightly heavier in the fall months
because of a desire to stockpile fuel for the region's long winters.
This comparison between the iron, grain, and coal industries

serves to briefly outline their major differences as seen in the
industrial histories of the three commodity trades. These differences
have an impact on a wide array of factors--the relationships between
which are not currently understood. From this historical sketch, a
number of elements appear of potential interest to the question of
representativeness. These factors are:

1) trade route locations

2) geographic variations along trade routes

3) commodity types (cargoes) transported

4) chronological development

5) vessel type and size

6) susceptibility to loss

7) ports of origin and destination

8) frequency of traffic

9) periodicity of navigational season

81

In the following chapter, these elements and others will be utilized to
form'hypotheses about the relationship between the development of the
shipping industry and the resulting shipwrecks. The model presented
will order these variables into a cohesive framework for study and

will suggest relationships between variables based on the historical

information discussed thus far.

CHAPTER IV
THE FORMATIVE PROCESS

This chapter organizes the historical information from the pre-
ceding chapters into a format that illustrates the interrelationships
between shipping and the formation of the shipwreck archaeological
record. The structure of the shipping industry was a result of numerous
factors linked to industrial expansion and development in several major
industries. Such factors as the physical environment, chronology of
development, degree of capital inputs in vessel technology, and the
seasonality of shipping conditioned the transportation of iron, grain,
and coal.

In regard to shipping, each of these commodities was structured
slightly differently because of their individual organization and
characteristics. Therefore, Lake Superior shipwrecks should reflect
the shipping industry as a whole as well as the specific commodities
that were transported across the region.

The shipping industry was a dynamic, constantly changing
system. As outside demand for commodities rose, so did vessel traffic.
And as economic decisions were made by individual industries or
companies, the composition of the vessel fleets in service on the lakes
changed accordingly. Theoretically, this process of change and develop-
ment should be reflected in the frequencies and types of shipwrecks that

occurred. As previously stated, the goals of this study are to search

82

83

for spatial regularities in the distribution of shipwrecks and to pro-
vide explanations for wreck patterning, if it is found to exist. This
can be accomplished in part through an understanding of the process by
which the shipwreck archaeological record was formed.

The formative process of the archaeological record is relevant
in several ways to the investigation of spatial distributions. First,
in order to recognize Spatial patterning of shipwrecks, it is necessary
to know and understand the patterning of the cultural context which
produced them. The shipping industry as a whole was structured in
several ways because of chronological development, trade routes, and
economic demand. Likewise, weather patterns, fog zones, and areas of
restricted navigation structured the physical environment. Both factors
had a major effect on the extent and development of the shipping
industry on Lake Superior.

Within the constraints imposed by the cultural and natural
environments, the shipping industry develOped into a major transportation
network responsible for the movement of large amounts of commodities and
people across the region. The iron, grain, and coal trades that serve
as the focus for this study were each affected a bit differently during
their development. For example, the iron industry had an earlier period
of development and had a more extensive movement of cargo across the
lake. In addition, the financial resources of the iron industry were
more substantial than either the grain or coal trades because of invest-
ment by wealthy eastern backers and later by huge multi-national
companies such as U.S. Steel. This allowed for more capital improvements
in terms of larger and more modern vessels; the iron industry fleets were

the most extensive in the Great Lakes region. This cultural context

84

provides a setting for the shipwrecks in the iron, coal, and grain
trades. This economic system, which serves as a backdrop for the study
of Lake Superior shipwrecks, has been referred to as the "systemic
context" by Schiffer (1972) and others (Reid, Schiffer, Neff 1975:210).
In contrast, the actual archaeological record (such as shipwrecks) is
referred to as the "archaeological context." These concepts are useful
not for the jargon they introduce but for the idea that there is a pro-
cess of transformation that takes place between these two contexts. The
formative process of the archaeological record is the means by which the
"systemic context" (the shipping industry on Lake Superior) is changed
into the "archaeological context" (Lake Superior shipwrecks). An
understanding of any aspect of archaeological sites depends on the know-
ledge of this formative process. In this study, it is useful to point
out that shipwrecks offer an opportunity for an elaboration of this
process because of the type of data available for study. The cultural
and environmental variables that affect wreck deposition are of two
types: 1) those variables contributing to loss; and 2) those relating
to the eventual decomposition of wrecks underwater. The present
condition of the archaeological record is therefore a result of these
two separate types of factors, which Schiffer (1972) has termed
"transforms." But Schiffer fails to differentiate between factors that
affect the initial deposition of a cultural phenomenon and those that
affect the extent to which decomposition takes place. For the purpose
of this study, these can be termed depositional transforms (Type A) and
archaeological transforms (Type B).

Type A variables are those factors that aid in the deposition of

cultural remains, while Type B variables relate to those factors that

85

aid either the preservation or decomposition of those remains. For
example, once an artifact is produced there are numerous factors that
influence how it is used and how it is eventually deposited ianitg,
(Type A variables). Some of those factors are cultural, such as the
economic system which dictates how that object will be used, while
others are noncultural, like the natural environment. Through a come
bination of factors, that object is eventually removed from the cultural
system either through intent or through loss. When the object is
deposited, it becomes a potential contributor to the archaeological
record. And the moment that deposition occurs, the artifact begins the
process of decomposition. At this point, Type B variables exert
influence on the object in question. Natural factors of decay affect
the object to some extent (this, of course, varies, depending upon the
object) either by chemical changes or by eroding or covering the arti-
fact. Cultural factors also come into play at that point, whether the
culture is the same that accounted for the formation and deposition or
whether it is a different culture in a different point in time. Some
cultural variables that may affect the artifact include land use, which
may disturb or destroy the object, and salvage, in the case of ship—
wrecks. The process of decomposition is ongoing and only ceases when
the original object is ultimately destroyed. Figure 7 illustrates this
formative process as it relates to Lake Superior shipwrecks.

In prehistorical archaeology, the primary context available to
archaeologists is the archaeological context. The other two, i.e.,
depositional and systemic, can only be inferred from an analysis of
archaeological materials or through such means as ethnoarchaeology and

experimental archaeology. On the other hand, historical archaeologists

86

someone o>Humauom one one exoouBeHSm .u oustm

 

 

mahowmnmua masowmooua
HmonoHoomnou< HosOHuHmooon
m mama < omza

    

     

 
  
     

   
  

   
 

      

    
   

 

 

ouooom mucous who on ueoauoHo>on
HmOHmoHooonou< when M axoouamHnm sumo w HmHuumooaH
usonoum H H o HoHuoosm mama H H o \womeeHAm
k1 \
A AA
uxouaoo nuouomw uxouooo muouomm uuouooo

   

   

 
 

HonouHoolsoz

 

HonouHsolooz HmsoHuHmOQoo oaaoumhm

  

HmonoHoomnou<

  

  

 

 

87

have the benefit of historical documents, which enhance the knowledge

available about the other two contexts and aid in uncovering the factors

responsible for the transformation process.

shipwreck studies of Lake Superior.

This is also true for

The historical documents available

on the region provide information concerning many of the significant

factors that influenced the production, deposition, and decomposition of

historic period lake vessels.

Both Type A and B transforms are discussed

in the literature (although to varying extents) so that many of the

factors responsible for the formation of the archaeological record are

known.

in this study.

Table 10.

CULTURAL

NON-
CULTURAL

Type A and Type B Factors

Type A
Depositional
Transforms

 

Economic strategies
Shipping behavior

Transportation routes CULTURAL
Commodity trades

Transportation technology

Weather-storms, fog, winds
Geography-water depths, NON-

harbors,
navigation hazards
Natural resource distribution

CULTURAL

Table 10 presents some of those factors that will be considered

Type B
Archaeological
Transitions

Salvage capabilities

Salvage decisions

Dredging operations

Sport diving

Legal protections

Dumping

Condition of wreck

Dollar value of
wreck

Historical value

Water depth

Ice damage
Wave actions
Water chemistry
Lake hydrology

The factors presented in this table are by no means a definitive

listing of all variables affecting shipwrecks.

Rather, it is illustra-

tive of the factor types that combine to form the underwater

88

archaeological record we now have. The shipping industry was a complex
system of relationships involving economics, technology, and culture
change. Likewise, the shipwrecks are reflections of these complexities
as well as of numerous other factors, such as the physical environment.

Discussion of the formative process of the shipwreck record
orders a wide range of variables into a unified system of thought. For
analysis, all the factors that influenced vessel loss and spatial
patterning can be grouped into two basic categories. The Type A
depositional transforms are those factors that contribute to vessel
loss, while Type B archaeological transforms affect vessels after loss
has occurred. These concepts are useful primarily as heuristic devices
for ordering and discussing the causal factors behind vessel loss and
distribution. .

Given the difference between the iron, grain, and coal trades
on Lake Superior, it can be assumed that each industry would be affected
differently by the factors outlined in Table 10. Theoretically, each
industry will undergo a different formative process in the deposition
and decomposition of related vessels, and the cultural and environmental
variables responsible for the formative process will be different for
each industry. Iron industry vessels, for example, may be differentially
affected by the environment (weather, fog, etc.) because the cultural
variables responsible for the transportation of iron (capital investment,
decision making, chronology) are different from.those of either the
grain or coal trades. Figure 8 illustrates the combination of factors
that result in vessel distributions. The interrelationships between

cultural and environmental factors are such that as cultural factors

89

 

 

swoops: swam
common weHoeHem
semen your:

season
soHumuuoeudsuu mo anemuwooo
onHa on huHHHoHquomom
masons ossuueuHs on %UHHHA<

muHm<Hm<> A<HZMZZOMH>ZM

 

mmooouo o>Humahou ecu eH moHomHum>

 

hquoaaou
.3
saoHusoHuumHn

Houmo>

N7

ouooom HoonoHoossoue

 

 

 

mo mmoooum o>Humauom

 

 

 

 

 

 

quooaaoo mom panama
usoadoHo>oo HsOHHOumHm
womxma.oonHoon
eHwHuo mo muuom

musooH HmuHeso

saws one some Hommo>

mmHdem¢> H4¢DHHDU

 

 

.m ouome

 

90

change, the formative process as a whole changes and different vessel
distributions result.

With this model of the formative process as a basis for the
remainder of this study, a series of hypotheses can be presented that
relates to spatial distribution of vessels. The following 21 hypotheses
test relationships between variables that apply to the question of
' spatial patterning. The first 15 hypotheses deal with the depositional
components (in this case--salvage) of vessels after deposition takes
place. Collectively, these hypotheses will probe the relationships
between key variables in the formative process as they relate to the

mechanism of spatial distribution and patterning.

Hypothesis 1
The iron, grain, and coal industries operated under different
procurement, distribution, transportation, and marketing conditions.
The related commodities that were shipped across Lake Superior were
transported along different trade routes with varying traffic frequency.
Because of these and associated factors, it would be expected that

there will be significant distributional differences between shipwrecks

 

of these three different commodity trades.
Test implications for hypothesis 1:

1) There will be significant differences between the distribution
of iron industry related shipwrecks and grain trade related
shipwrecks for the period 1855-1920.

2) There will be statistically significant differences between the
spatial distribution of iron industry related shipwrecks and coal

trade related shipwrecks for the period 1855-1920.

91

3) There will be significant differences between the spatial
distribution of grain trade related shipwrecks and coal trade

related shipwrecks for the period 1855-1920.

Hypothesis 2

As the development of the Lake Superior region progressed and
new water transportation routes were established, the flow of traffic
across the lake was markedly altered. For example, as the iron industry
expanded into new areas of production in the western portion of the
region, the routes of ore transportation went through a series of
modifications. Theoretically, corresponding shipwrecks would also have
changed distribution if they are representative of past behavioral
changes. Therefore, it would be expected that there will be significant
distributional differences of shipwrecks by chronological period.
Test implications for hypothesis 2:

1) There will be significant distributional differences of ship-
wrecks by chronological period for iron, grain, and coal related
vessels.

2) There will be significant distributional differences for all Lake

Superior shipwrecks by chronological period.

Hypothesis 3
Within the iron and grain industries, there were different but
related types of commodities traveling along similar routes of trans-
portation. The grain trade involved the shipment of wheat, oats,
barley, corn, and flour from the Duluth and Port Arthur ports. The iron
industry out of Marquette focused on both the local production and

shipment of pig iron as well as the movement of iron ore in the pre-l880

92

period. The particular qualities of these commodities may have resulted
in different patterns of loss for each cargo type. Therefore, it would
be expected that there will be significant distributional differences
between related cargg types within theygrain trade and iron industries.
Test implications for Hypothesis 3:
1) There will be significant distribution differences between lost
vessels carrying a) oats, b) wheat, c) corn, d) barley, and e)
flour from Duluth for the period 1870-1920.
2) There will be significant distribution differences between lost
vessels carrying pig iron and vessels carrying iron ore for the

period 1855-1880.

Hypothesis 4

 

Transportation on the Great Lakes is seasonal by nature and
restricted to the months of ice-free navigation. This of course varies
from year to year and is dependent upon the pecularities of weather in
the fall and spring months. When the navigation season is short because
of an early winter or late spring, more cargo must be moved in less time
than in a longer season. This results in greater traffic congestion on
the Lakes during the navigable months, as well as more risktaking by
companies and their vessel captains in order to insure that the maximum
amount of cargo will be moved. On Lake Superior, the length of the
shipping season was very important for the commodity trade in bulk
cargoes because of the rack of alternate economical means of transporta-
tion during the winter months. For the most part, grain, coal, and iron
ore were very dependent upon water transport for marketing and distri-
bution. The combination of congested traffic and a need for maximized

cargo movement may have had an effect on vessel loss. For these

93

reasons, it would be expected that there will be an inverse correlation

 

between the length of the nayigation season and the frequency with which
vessels of different commodities were lost.
Test implication for Hypothesis 4:
1) More grain, ore, and coal shipwrecks will occur during years
with short navigation seasons than with average or longer

seasons .

Hypothesis 5
Throughout the history of the Great Lakes, weather has been a

significant factor in vessel loss. On Lake Superior, the occurrence of
storms, fogs, and other hazardous conditions has led to the demise of
many vessels. Storms, for example, are particularly prevalent from
mid-October to the close of navigation, while fog is of greater concern
during the early months of the seasons when differences between water
temperature and air temperature are condusive to fog-producing
conditions. These conditions are of importance to all vessels on the
lake regardless of the specific cargo being carried. Since weather is
a natural factor, it should therefore effect all vessels carrying grain
or coal to the same extent as those carrying ore. It would be expected
that there will be nogsignificant differences in the frequency of
stormrrelated shipwrecks between vessels with caggoes of specific
commodityitypes.
Test implication for Hypothesis 5:

1) Stormrrelated losses of vessels in the iron industry will be the

same as losses in the coal or grain trades.

94

Hypothesis 6

The transportation routes followed by Lake Superior vessels were
different in a number of ways. The distance of routes varied as did the
navigational hazards encountered along those routes. These factors
combined to make some routes more dangerous than others. Ports of
origin for lost vessels provide information on the routes being
traveled at different points in time. Since the port of origin is a
reflection of the transportation route, and because all routes are both
geographically and functionally different, it would be expected that
there will be distributional differences between vessels of different

ports of origin.

 

Test implication for Hypothesis 6:
1) Lost vessels from Duluth, Two Harbors, Ashland, Marquette, and

Port Arthur will exhibit different patterns of distribution.

,Hypothesis 7

The commodities of iron, coal, and grain operated under differ-
ent procurement, transportation, distribution, and marketing systems.
These differences may have resulted in corresponding differences in the
types of vessels employed in the transportation of these commodities.
Despite the fact that all three commodities were transported by bulk
carriers, there may be some differences in vessel types. For example,
because of the relatively greater importance of iron ore over grain,
there may be differences in the types and sizes of vessels selected to
carry each commodity. Iron ore vessels may be larger in average size
than vessels of the grain trade for each respective period of time.

Therefore, it would be expected that there will be statistically

 

significant differences between the types of vessels employed in the

95

iron trade as opposed to either thqurain or coal trades.
Test implications for Hypothesis 7:
1) There will be statistically significant differences in vessel
types lost with cargoes of ore, as opposed to carriers of grain

or coal.

Hypothesis 8
Because of the economics of water transportation, there was a

steady increase over time in the size of lake vessels specializing in
the movement of bulk cargoes such as iron ore, grain, and coal. The
trend toward larger vessels required modifications in vessel design and
structure, resulting in the gradual development of new types of vessels.
The change from schooners, to propellors, to schooner-barges, and
finally to steamers is well-documented for the Great Lakes, including
the Lake Superior region. It would therefore be expected £322
shipwrecks will reflect the changes in vessel type that occurred over
IQEE?
Test implication for Hypothesis 8:

1) The types of vessels lost will vary in frequency by the

chronological period of development.

Hypothesis 9

 

Lake Superior shipwrecks can be categorized into four major
causal types of loss: 1) collision, 2) grounding, 3) fire, and 4)
foundering. These types of loss were a result of a combination of
natural and cultural factors affecting vessels in different ways.
Because of the differences in vessel type, transportation routes,

cargoes, and relative frequency of traffic, these types of loss may have

96

had greater influence on vessels of some industries than others. For
example, the nature of grain cargoes may have a bearing on the types of
loss associated with grain trade vessels. Therefore, it would be
expected that there will be significant differences in the extent to
which each commodity (i.e., coalipgraini iron) was affected by each of
the four major types of loss categggies.
Test implications for Hypothesis 9:
1) Fire will have affected vessels of the iron, coal, and grain
trades to a significantly different extent.
2) Collision will have affected vessels of the iron, coal, and
grain trades to a significantly different extent.
3) Foundering will have affected vessels of the iron, coal, and
grain trade8.to a significantly different extent.
4) Grounding will have affected vessels of the iron, coal, and

grain trades to a significantly different extent.

Hypotheses 10 and 11
The four types of vessel loss (i.e., collision, grounding,

foundering, and fire) are different phenomena with different causal
factors contributing to vessel 1088. Those factors behind type of loss
include transportation routes, navigational hazards, transportation
technology, weather, etc. Because these same factors contribute to the
distribution of shipwrecks in Lake Superior, there may be a link
between the type of loss affecting a vessel and the location of the

lost vessel. Therefore, it would be expected that there will be

 

significant distributional differences between vessels of different

loss categories.

97

Test implications for Hypothesis 10:
1) There will be significant distributional differences between
vessels lost by collision, grounding, foundering, and fire.
In addition, since the type of loss is correlated with transportation
routes and transportation technology that change through time, it
would also be expected that the frequencies of vessels in each of the

four loss categpries will vary by chronological period.

Hypotheses 12 and 13

As vessel design changed through time, the ability of those
vessels to cope with the hazards of the regions also changed. For
example, the change from schooners to steamrpowered craft decreased the
length of travel from one point to another, providing greater indepen-
dence from the unpredictability of winds and weather. To some extent,
the new steam craft could better cOpe with the environment and would
therefore be less prone to losses caused by winds or weather. But,
the physical structure of the new steam vessels might have made them
more vulnerable to other hazards. It would therefore be expected that
there will be a correlation between some vessel types and corregponding

types of 1088..

 

Test implication for Hypothesis 12:

1) Vessel types such as schooners, schooner barges, propellers,
wooden steamers, and iron steamers will correlate more with
certain types of vessel loss than with others.

In addition, because some geographic areas may have placed more
stress upon some vessel types than other areas, it would also be expected
that there will be significant distributional differences between vessels

of different types.

98

Test implication for Hypothesis 13:
1) Schooners, schooner-barges, propellers, wooden steamers, and iron
steamers will exhibit significant differences in distributional

patterns.

Hypothesis 14

 

The iron, grain, and coal commodities transported on Lake
Superior varied in importance. Iron ore, for example, was transported
in much greater quantities than either grain or coal. The relative
importance of these commodities had a major effect on their transporta-
tion as well as on the frequency of traffic associated with each. If
frequency of traffic is related to frequency of loss, then it would be
expected that there will be a direct correlation between the relative
freqpency of traffic and the correspondingfifrequency of vessel loss.
Test implication for Hypothesis 14:

1) There will be a direct correlation between the frequency of iron
industry traffic and the frequency of vessels lost in that

industry.

Hypothesis 15

Vessels of the Lake Superior shipping industry traveled along
different transportation routes because of the origins of the various
'commodities. The iron industry shipments originated from the ports of
Two Harbors, Ashland, Duluth, and Marquette, while the grain trade
operated primarily from.Duluth and Port Arthur. The frequency of
traffic from these ports in the associated commodities should therefore
be reflected in the relative origins of the vessels lost in the

corresponding chronological periods. For example, if 50 percent of the

99

grain shipments for the period 1870-l910 originated from Duluth, then
50 percent of the associated grain vessels lost should also have
originated from that port. Therefore, it would be expected that £2352
will be a direct correlation between the frequency of traffic in a_given
commodity from a given port and the associated frequency of vessels lost

from that_port.

 

Test implication for Hypothesis 15:
1) The ratio of iron industry vessels lost from Two Harbors,
Ashland, Duluth, and Marquette will directly correlate with the

frequency of traffic from those ports for the period 1855-1920.

Type B Archaeological Transform Hypothesis

Archaeological transforms are those cultural and environmental
factors that influenced the formation of the archaeological record from
the point of deposition to the present condition of the record.
Although there are numerous factors that affect shipwrecks in Lake
Superior, only those elements that are historically documented will be
considered here. As future underwater archaeological investigations in
the region are undertaken, increased knowledge of these factors will
accumulate. Such studies will enhance our knowledge of archaeological
transforms and their influence on the formation of our present under-
water archaeological record.

Several hypotheses here relate to Type B transforms involving
the element of salvage. Salvage is an important factor to consider
because once vessels are deposited on shore or on the bottomlands, they
become a potential source of the archaeological record. If they are
consequently salvaged, they never become a part of the shipwreck record.

The influences of salvage on the archaeological record can be learned

100

in part by comparing those vessels that were not salvaged to those
vessels that were. Figure 9 illustrates the role of salvage in the
formation of the shipwreck archaeological record. In order to under-
stand a large portion of the mechanics of the Type B transformation
process, it is essential to understand the salvage behavior that conr
tributed to this process. Unfortunately, historical records involving
salvage are incomplete and difficult to obtain. This situation
necessitates that other measures be used to arrive at both the question
of the extent to which salvage was accomplished and whether the vessels
salvaged are spatially patterned.

If the locations and characteristics of salvaged shipwrecks are
the same as those that were not, then the question of patterning is
resolved. If, however, there are differences between the two groups
(salvaged/not salvaged), then it is necessary to evaluate the extent of
these differences as well as possible reasons for these differences.

So the major hypothesis for this section is that if the depositional
context is spatially patterned, then the archaeological context will be
patterned in a like manner.

An investigation of this major hypothesis, as well as of the
special conditions of salvage, will be outlined by the following

hypotheses.

Hypotheses 16-21
Because of the factors discussed in this last section, we can
hypothesize that if no relationship exists between certain factors and
the salvagability of vessels then:
Hypothesis 16-There will be no significant differences between

salvaged vessels and nonsalvaged vessels in regard

101

Figure 9. Salvage and the formative process

 

102

muouosw Hoeuo
nwsounu mnoH

    

unsusoo
HsOHonoossoue

    

 

  

somuoHseoe
soouseHeo
usououm

      

ows>Hsm

 

unousoo
HmaoHOHmoeon

oomp>Hmm

 

 

  

nos oxoouseHnm

 
 

omoH hHHmusHuo
axooHanem

mo eoeuuHsoom

  
  

mo dowumHamom

1&1

assouuaaue m «use

Hypothesis

Hypothesis

Hypothesis

Hypothesis

Hypothesis

103

to vessel type.

17-There will be no significant differences between
salvaged vessels and nonsalvaged vessels in regard
to the port of origin of the respective vessels.

18--There will be no significant differences between
salvaged vessels and nonsalvaged vessels in regard
to the month of loss.

19--There will be no significant distributional differ-
ences between salvaged vessels and nonsalvaged
vessels for each commodity and for all vessels lost
on Lake Superior.

20--There will be no significant differences between
salvaged vessels and nonsalvaged vessels in regard
to the type of loss that brought about the
respective shipwrecks.

21-There will be no significant differences in regard
to the frequency of salvage between vessels of the

iron, grain, and coal trades.

CHAPTER V
METHODOLOGY

This chapter outlines and discusses the methods by which the
hypotheses posed in chapter 4 will be tested and evaluated. Of the 21
hypotheses presented, as many as possible will be tested, given the
limitations of the data base. In chapter 6, each individual hypothesis
will be restated and tested using the techniques set forth in this

chapter.

Shipwreck Attributes Defined
In order to test hypotheses concerning Lake Superior shipwrecks,

it is necessary to define the types of data needed to complete the
analysis. As previously discussed, shipwrecks are extremely diverse in
their individual characteristics because of their chronology, function,
or deposition. Despite this diversity, a number of attributes are
common to all shipwrecks and serve to cross-cut the population of Great
Lakes vessels. These attributes form the basis for the hypotheses pre-
sented in chapter 4 and will serve as the data base from which hypotheses
can be analyzed in the following sections. Following is the list of key
attributes that will be dealt with throughout the remainder of this
study:

1) Shipwreck category - total loss vs. salvaged vessels;

2) Vessel type - schooner, schooner-barge, wooden steamer, steel

steamer, misc.;

104

105

3) Date of loss - monthiand'year;
4) Point of origin - Duluth, Two Harbors, Ashland, Marquette,
Port Arthur, misc.;

5) Location of loss - in Poisson grid;

6) Cargo - iron, grain, coal;

7) Type of loss - fire, grounding, foundering, collision;

8) Weather conditions at time of loss - presence/absence of storms,

fog.
Other attributes not related specifically to shipwrecks but

rather to the shipping industry as a whole include:

9) Frequency of traffic from ports;

10) Routes of travel;

11) Frequency of loss;

12) Frequency of salvage.
These 12 attributes are essential to the successful testing of the
hypotheses under consideration as well as the interpretation of the
test results.

Abstraction of Attributes from Historical
Sources
After the essential attributes needed to test hypotheses are

defined, the data-gathering phase of the research can begin. In
relation to this study, the body of information from which attribute
data can be abstracted consists of several historical treatments of
Great Lakes shipwrecks. These studies are either historical case
studies on the losses of individual vessels, such as Wolff's (1979)
recent work, or they are inventories of lost vessels compiled by

various authors (Heden 1966; Winkleman 1971). These sources provide an

106

information base that deals with vessel attributes in either a direct
or indirect manner. Wolff's (1979) study in particular was a detailed
accounting of more than 1,000 Lake Superior accidents. These case
studies have been abstracted for details on vessel attributes, and data
sheets on each lost vessel have been prepared. As a base for this
study, information on a total of 146 recorded shipwrecks has been
abstracted from historical sources. The data sheets compiled for these
vessels constitute a collection of vessel attributes that can be drawn
upon and quantified to test the hypotheses.

A total of 146 lost vessels is relatively small in relation to
the actual number of accidents that have occurred over the years, but
it is the total population of known recorded losses for the period.
Although there may be a few vessel losses that were not recorded for
various reasons, given the reliability of the historical sources used,
there is no reason to believe that many, if any, losses went unrecorded.
The one limitation of the data is that the list of 146 vessels is small
and that in approximately 20 percent of the cases, data on individual
vessel attributes are partial. For example, case studies on some
vessels fail to list one or more characteristics, such as point of
origin or condition of loss, and are therefore not as complete as could
be desired. But since most of the case studies allow for the full
completion of data sheets, this does not appear to be a major liability.
The lost vessels and their associated attributes are listed in a cone

densed form in.Appendix B.

107

Quantification of Attributes

 

After data sheets are prepared from historical treatments of
shipwrecks, the attributes listed on the sheets can be tabulated by
category. For example, the type of vessel loss is of relevance to the
testing of several hypotheses. Fire, grounding, foundering and collis-
ion are the four categories of loss into which most (if not all) types
of vessel losses can be placed. From the data sheets, a simple count
can be made of all vessels of known loss types and the relative fre-
quencies can be tabulated for each of the loss categories. With a
sample size of 146, all counts could be completed by hand without the
use of computer assistance. Finer breakdowns of attributes can be made
as desired so that, for example, the relative frequencies of loss types
for both salvaged and totally lost vessels could also be tabulated if
so desired. By this means, attributes can easily be cross-tabulated
and quantified for comparisons using statistical tests. The limitation
of a small sample size prevents some cross-tabulations from being used
because of the very low frequencies encountered when numerous categories
are used. A cross-tabulation between loss type and vessel type in
regard to those vessels lost in storms (for example) would prove to be
unfeasible because of partial data and low sample size. But most
cross-tabulations are easily completed, as are simple attribute counts
and presence/absence totals. The resulting raw scores of vessel or
attribute counts are then analyzed using statistical techniques

appropriate for nominal scale data.

Statistical Evaluation

 

The frequency counts of vessel attributes obtained from the

data sheets are then evaluated through the use of several statistical

108

techniques. Those tests used in this study are 1) Poisson distribution
for locational data, 2) Kolmogorov-Smirnov test for goodness of fit
(used in conjunction with Poisson), 3) Contingency Chi Square, and 4)
Kx2 Chi Square; the last two tests are used to test strength of
associations between attributes. These will be briefly described here,

and chapter 6 contains examples of each.

Poisson Distribution

The Poisson distribution is a frequency distribution of rare
events that is randomly patterned. When the Poisson distribution is
compared to a distribution derived from a sample group, it can be
determined whether or not that sample is likewise randomly patterned.
This technique has been commonly used in probability mathematics and
statistics since its first description in 1837 (Sokal and Rohlf 1969:
84), and in recent years, it has been applied to point pattern analysis
by geographers and archaeologists (Hodder and Orton 1976:34). This
latter application of Poisson is accomplished by means of a grid system
imposed over a spatial plane and the tabulation of a variable's
frequency of occurrence for each cell of the grid. The frequencies of
variables occurring throughout the cells of the grid area are then
compared with an expected (Poisson) distribution of those variables to
determine if the variable is randomly or nonrandomly spatially patterned.

The use of Poisson in this study is particularly well-adapted
to the determination of patterning among Lake Superior shipwrecks. By
definition, shipwrecks are rare events and relatively scarce in
relation to the tremendous amount of traffic on the lake in a given
season. So the use of Poisson to evaluate spatial patterning of wrecks

is very appropriate to small samples (Sokal and Rohlf 1969:81-95). And

109

since only 146 vessels are recorded lost on Lake Superior for the perio”
in question, thisalso appears to be an appropriate test of the data.
For the use of Poisson on Lake Superior shipwrecks, a grid
system was established across the lake based on latitude and longitude
in 30~minute increments. The resulting grid system was composed of
65 rectangular cells of equal size (see Figure 10). Although
theoretically these cells were of equal area, in actuality some cells
were smaller than others because they fell on land areas that would not
afford equal access as water arenas. For example, cell 58 on Whitefish
Point is roughly half water and half land in area. This is in contrast
to cell 42, immediately to the north, in which the surface area is 100
percent water. Theoretically, this differential cell size could have
an impact on the frequency of vessels lost in the respective cells.
Although this problem was carefully considered and adjustments in
scores (or cells) could have been made to correct for such differential'
access, it was decided to leave the grid uncorrected. This decision
was based on a cursory examination of wreck distributions to determine
if cell counts were substantially higher for all-water cells versus
partial-water cells. The results of this examination showed that
shoreline cells were in actuality significantly more associated with
wrecks than all-water cells. For example, cell 58, which was 50 percent
water, had a total of 15 losses, while cell 42, which was 100 percent
water, had no wrecks for the same period. It appears that although
there may be some differential access across cells, there is no major
block to patterning. If correction techniques were used, the results
of the Poisson comparisons would be intensified rather than smoothed.

Since this intensification is probably not necessary, the decision was

Figure 10.

Poisson grid system

110

111

 

V
maiz up

h(.—m

1N

 

u.¢(

 

.whm

p :(m

 

02.2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F9

 

 

 

 

 

 

ion/3

 

 

 

 

 

—T~

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

112

made to accept the 65-cell grid system despite its imperfections.

Once the grid system.was imposed across the lake surface, the
locations of wrecks were plotted based on the information provided in
the historical sources. In approximately 75 percent of the cases, the
location of loss was very exact-especially with groundings. For
example, a location listed as a grounding one mile east of the Big Two
Hearted River (cell 58) is easily plotted on the map and little
difficulty is encountered assigning the loss to a grid cell. In other
cases, a location listed meiles west of a certain point is less easily
plotted on the map and is less exacting. There is little problem
assigning these locations to a grid cell despite the lack of exact
location because each grid cell is approximately 820 square miles in
area. With the use of a grid system, if a location is mis-plotted by
a couple of miles one way or the other, there is usually no problem
placing vessels into incorrect cells. But if the number of cells was
increased and the square area of each therefore decreased, some problems
may occur. The 65-cell grid system used for this study was large
enough to facilitate plotting of vessels yet small enough to detect
subtle variations in patterning that would otherwise be masked with
larger cells. In only a few cases was vessel location difficult to
plot, and when such instances arose the plot was either done judgmental-
ly if a reasonable assurance of accuracy could be made, or was not
plotted if locational information was deemed insufficient. A location
listed as "between Keweenaw Point and Whitefish Point" was not plotted
because it lacked precise information. Of 146 total shipwrecks, only
13 (9 percent) could not reliably be plotted or did not have locational

information provided.

113

Once vessels were located on the map and assigned a cell number,
tabulations could be made in relation to particular attributes. For
example, all vessels from the Marquette port could be tabulated by their
frequency of occurrence in the grid cells. Some cells had no vessels
from Marquette, while others had one or more. The frequencies of the
Marquette vessel distributions in these cells could then be compared to
an expected random.Poisson distribution for determination of
patterning. Vessels from.Marquette could then be determined as being
randomly distributed across the lake or nonrandomly distributed in
either a "regular" or "clumped" pattern. The actual type of pattern
can be determined by the computation of the coefficient of dispersion
(CD), which is equal to the variance (82) divided by the mean (§).

This coefficient will be near 1 in distributions that are essentially
Poisson, greater than 1 in clumped samples, and less than 1 in regular
frequency distributions (Sokal and Rohlf 1969:88; Hodder and Orton
1976:34). In addition, since the variance is a function of the mean in
a Poisson distribution, the variance will equal the mean when the dis-
tribution is randomly patterned.

The following formulae were used to calculate the relative
expected frequencies of the Poisson distribution.

11y FYZF $33?

if.
’ _ ’ _ ’ _ 9 _ , o o 0
Y
e

Y Y 1 2 ZeY 3 2 x 3eY

 

 

where T'- sample mean and #7 is the function of the mean. To obtain
the absolute expected frequencies, the first term must be multiplied

by n, the number of samples.

-2

 

 

f0_ :1- , flall 3?, f2. n: 1" f3_ nY_ 1,.
Y Y Y 2 Y 3
e e e 2e

The variance, 82, can be determined through the following formula:

Variance = Z fny2

n

where y - # per cell - mean and f 8 observed frequency.
The coefficient of dispersion is calculated in the following manner:

2
8

Y

 

C.D. -

The actual computation of Poisson distributions will not be
discussed in this study since it is a well-known technique.. The reader
is directed to Sokal and Rohlf (1969:81-95), Steel and Torrie (1960:
395-399), Doran and Hodson (1975:44-51), Hodder and Orton (1976:33-38),
and Dacey (1968:172-180) for specifics on computation and applicability
of Poisson distributions. Generally, however, the computation of
Poisson is accomplished through the comparison of observed frequencies
(such.as the number of shipwrecks occurring in each.of the 65 cells of
the Lake Superior grid system) to computed expected frequencies that
would be randomly distributed based on the mean and sample size. A
goodness of fit test is then computed on the deviation between the
observed and expected values and a determination of association is
arrived at for various levels of significance. Both the Chi Square and
Kolmogorov-Smirnov tests are applicable to this regard (Hodder and

Orton 1976:38).

115

Kolmogorov-Smirnov Test of Goodness of Fit

In order to evaluate deviations between Poisson distributions
and observed distributions, a goodness of fit test is required. For
this purpose, the Kilmogorov-Smirnov test is especially well-suited
because it is not subjected to the limitations of sample size and
expected frequencies, as is the Chi Square (X2) test. With the X2 test,
the observed frequencies must remain above 5, otherwise the tail of the
distribution becomes distorted, affecting X2 interpretation. This
often necessitates the merging of groups to obtain the needed frequency,
which in turn reduces the degrees of freedom.and may obscure the
deviations from.the predicted Poisson distribution (Hodder and Orton
1976:38). With the Kolmogorov-Smirnov test, cumulative observed and
expected frequencies are computed, as are cumulative deviations. This
cumulative process eliminates the need for merging groups and is more
reliable for testing small samples, such as is the case in this ship-
wreck study. The point of maximum deviation is then divided by the
sample number (which is 65 because of the grid), and the statistic D is
obtained. This statistic is then compared with critical values to
determine confidence levels for accepting or rejecting the null hypo-
thesis that the observed frequency distribution fits the Poisson
(random) distribution. It is necessary to calculate the cumulative
observed, F, and cumulative expected frequencies, F,-the8e frequencies
are subtracted to determine cumulative deviations and ultimately
identify dmax' Thus d - F - F and the largest valu: is labeled
dmax' The following formula is used to find D; D --E%¥. For further
discussion of computation and applicability, see Siegel (1956:47-60)

and Sokal and Rohlf (1969:571-575).

116

Contingency Chi Square

The contingency chi square test measures the extent of
association or relation between two sets of attributes (Siegel 1956:
104-111, 175-179, 196:202; Sokal and Rohlf 1969:550-572). For the
purpose of this study, this test was used to test the association
between shipwreck attributes. Raw scores tabulated from data sheets
provided the nominal scale data for tabulations of observed frequencies.
Using two attributes, a contingency table is constructed of observed
frequencies. From these frequencies, the expected frequencies for each
cell are generated by multiplying the sum of rows by the sum.of columns
in the table and dividing by the total sample size. For example, an
analysis of the association between different commodities (iron, grain,
coal) and types of vessel loss (collision, fire, grounding, foundering)
would result in a 12-cell contingency table with observed frequencies
being the number of vessels in each of the industries that were wrecked
due to each of the four categories of vessel loss. The sums of fre-
quencies for loss type (rows) and commodity types (columns) would be
computed and multiplied by corresponding cells for division by the total
sample size. The Chi Square values for each cell would then be computed
by summing the square of the deviations between observed and expected

frequencies and dividing by the expected values.

 

In this manner, a X2 value for each cell is computed and the relative
association between attribute elements can be interpreted by the size

of this value. The final step of the process is to sum all the X2 values

117

in each cell and to compare that figure to X2 tables for confidence
evaluation. A total chi square value greater than the tabular value is
grounds for rejecting the null hypothesis that there is no association
between attributes (e.g., commodity and type of loss). This test is
used extensively in this study to evaluate the degree of association

between numerous combinations of vessel attributes.

Kx2 Chi Square

This test is mathematically similar to the contingency Chi
Square just described but has different application. The Kx2 test
provides a test of association between one attribute and two other
attributes, the latter of which are variations of a single population.
For example, the extent of association between type of vessel loss and
the differential degree of salvaged versus totally lost vessels can be
computed using the Kx2 Chi Square method. For the hypotheses that
involve the attribute of salvaged/total loss, this test can be used to
determine strength of association between the segments of that attribute
and other attributes such as vessel type and loss type.

The computation of the X2 value is accomplished through an
analysis of, for example, proportions between lost and salvaged
vessels relative to the sample size. See Steel and Torrie (1960:370-
371) for specifics of computation. The overall result of this computa-
tion process is a X2 value that can be evaluated for association at
confidence levels. The minimum acceptable confidence level for both
the contingency X2 and KxZ X2 values is the .05 level. A computed X2
value greater than the confidence value is grounds for rejecting the

null hypothesis that the ratio of salvaged to nonsalvaged (total loss)

118

vessels does not vary by more than chance from one attribute element
(e.g., collision, fire, grounding, etc.) to another.

These four tests (Poisson, Kilmogorov-Smirnov, Contingency X2,
Kx2 X2) form.the basis for the statistical evaluation of shipwreck
attributes used in this study. In the following chapter examples of

these tests will be provided as each hypothesis is analyzed.

Interpretation of Results

As the logical conclusions to the testing of hypotheses pre-
sented in chapter 4, chapter 7 will synthesize and interpret results.
As a methodological step in this research, this will be the most
important step toward evaluating the question of representativeness,
around which this study is based. Since the subject will be treated
in detail, it is unnecessary to describe here the specifics of that
evaluation process. It is useful to say that the relationship between
attributes are hierarchically ordered so that the results of hypotheses
testing of earlier propositions will have a bearing on the interpreta-
tion of later prOpositions. For example, hypothesis 1 proposes that
distributional differences will occur between shipwrecks associated
with the iron, grain, and coal industries. The results of the testing
of hypothesis 1 will then have a bearing on the interpretation of re-
sults (not the results themselves) of other hypotheses that deal with
both distributional differences and commodity differences. Chapter 7
will describe in detail the process of interpretation and will discuss

the specific links between hypotheses.

CHAPTER VI

HYPOTHESIS TESTING

Introduction

 

The hypotheses tested in this chapter involve the formative
process of the archaeological record as it applies to the spatial dis-
tribution of irons, grains, and coal-affiliated shipwrecks. Central to
the question of spatial patterning are the individual variables that
contribute to vessel loss, deposition, and decomposition, as well as
those factors that influence the shipping industry as a whole. The 21
hypotheses tested here have been posed because of their relevance to
the understanding of the patterning process. As previously discussed,
the first 15 hypotheses relate specifically to the process of vessel
loss, while the remaining six discuss the role of salvage in the
current deposition of vessels.

It is important to outline the relationships between the
variables being tested as well as the specific hypotheses used to
accomplish the tests. The variables examined in this chapter are
closely interrelated and often have an impact on one another through
direct or indirect means. These variables can be viewed as a systems
framework that links each element together into a cohesive network. To
establish the association between the variables, 21 hypotheses were
created to explore these relationships as they apply ultimately to the
spatial distribution of shipwrecks in Lake Superior. Figure 11

graphically illustrates the variables under consideration along with

119

120

Figure 11. Variables under consideration

121

 

 

mark

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmOH
m. m—
ex S
cm
a
.—
mmwa monH<UOH
ammmm> mu<>q<m MommzmHMm
o. a.
one: a? «x uuom
.N
e/
.6
a
N
HHHoozzoo
Hsoosomh m _1uoeuoo31_ o~ QHHHauH_

 

122

the numbers of theEspecific hypotheses designed to test the associations
between these variables. As this diagram.shows, ten major variables
are dealt with in this study. These variables are characteristics of
shipwrecks that are documented in the historical record and that are of
use to the understanding of the shipwreck phenomena. These are: 1) come
modity affiliation (iron, grain, coal); 2) frequency of traffic; 3)
weather; 4) seasonality of shipping; 5) location of lost vessels;
6) port of origin; 7) degree of salvage; 8) date of loss; 9) vessel
type; and 10) type of loss. These characteristics of shipwrecks are
used in the 21 hypotheses presented in chapter 4. Lines connecting
each of the shipwreck variables illustrate the presence of one or more
hypotheses used to test the relationship between these variables. The
hypotheses numbers shown on the diagram correlate with the hypotheses
tested in this chapter. For example, hypothesis 1 tests the relationr
ship between commodity affiliation (iron, grain, coal) and the spatial
patterning of Lake Superior shipwrecks. The function of this diagram
is to illustrate the role of each hypothesis in linking two or more
variables for study. As Figure 10 shows, there are many variables that
are not directly linked to one another and that will not be tested. The
hypotheses presented are therefore only testing those relationships that
are pertinent to the question of patterning and that can be tested by
the available data. Many questions will be left unasked in this study
because of a lack of data or relevance to the issue at hand. HOpefully,
future studies using a larger data base or a different topical emphasis
will address some of the relationships not explored in this research.
Hypothesis 1-15 examine the variables that contribute to vessel

loss and deposition (Type A transforms), while hypotheses 16-21 explore

123

the factors, such as salvage, that contribute to the process of decomr
position (Type B transforms). Within the outline provided, the role of
each hypothesis can be viewed as a mechanism for testing the relation-
ships between the variables presented. As Figure 12 shows, it is
obvious that some variables play a more important role in testing the
relationship to shipwreck locations than others. The elements of
commodity affiliation, vessel type, loss type, and salvage are of
particular importance to an understanding of the ultimate spatial
distribution of shipwrecks. Such characteristics as weather, seasonal-
ity, and frequency of traffic are also important to the investigation
of wreck distributions but in a more indirect manner. These three
variables are tested through their influence on individual commodities,
which in turn can be related to the question of spatial patterning. In
this way, all variable relationships can be tested as they relate to the
specific focus of this study.

Another way to view these key variables is through an elaboration
of Figure 8, presented in chapter 4. Because these variables have an
ultimate effect upon patterning of shipwrecks, their role in the forma-
tive process must be outlined. Figure 12 illustrates the variables that
affect the Spatial distribution of iron, grain, and coal related ship-
wrecks. The cultural factors of economic demand, transportation routes,
chronology of development, and technological change are reflected in
such variables as frequency of traffic, port of origin, date of loss,
and vessel type respectively; environmental variables of weather and
geography influence vessel loss due to storms, fog, navigational
hazards, etc.; the interaction of cultural and environmental variables

contributes to vessel loss and distribution of resulting shipwrecks.

124

Figure 12. Variables affecting iron, grain, coal vessels

125

 

 

monH<UOH MommzmHmm

UZHZMMHH<m A<HH<mm

 

 

 

 

swoops:
.mousuos
HssoHust>ms 1 honeymooo

common wsHooHsm
mo sumsoH
.wom .mauoum 1 scenes:

 

mm4m<Hm¢> A<Hzm=zcmH>zm

x7

mooooum o>Humanom

 

 

 

 

 

 

owsmno HoonoHossooH
mooH no more
usHuo mo uuom

onmmuu mo hososoonh

 

mMHn<Hm<> A<mDHHDU

 

 

 

126

’A

For each industry (iron, grain, coal), the formative process is differ-
ent because the cultural variables are different. The iron industry was
different from.either grain or coal because of numerous factors dis-
cussed in earlier chapters. The hypotheses that are tested in this
chapter seek to understand the effect of commodity differences in the
spatial patterning of shipwrecks. The question is “do iron industry
wrecks pattern in the same manner and with the same characteristics as
the other two commodities, and if so, why?" The answer to this question

and others will depend upon the results of this testing process.

Hypotheses

The following hypotheses deal with numerous factors involved in
the process of spatial patterning and deposition. These factors are
either cultural or environmental in nature and often involve combina-
tions of one or more elements. As a whole, the first section of
hypotheses represent Type A transfonms. Such factors as 1) distribu-
tional differences within and between industrial commodities, 2)
frequency of vessel traffic, 3) vessel origins and transportation
routes, 4) transportation technology, 5) period of navigation, 6)
weather patterns, and 7) types of vessel loss will be examined through
the formation and testing of related hypotheses.

The second section of hypotheses deals with Type B factors and
involves the cultural and noncultural influences upon the archaeological
record after deposition takes place. Because this study is primarily
based on the historical record, Type B transforms are necessarily
somewhat under emphasized. It is hoped that as future archaeological
endeavors probe the underwater shipwreck record, knowledge of these

factors can be greatly expanded. Until that time, the historical

127

record must suffice despite its potential inaccuracies and information

gaps.

Type A Depositional Transform Hypotheses

The overriding hypothesis concerning the transformation of the
systemic context into the depositional context is the degree to which
the latter is a patterned representation of the former. Re-stated,
this hypothesis postulates that shipwrecks originally deposited on the
bottomlands are spatially reflective of past patterned cultural
(shipping) systems, so patterns of vessel loss will be found that will
closely approximate the system from.which they developed. With this
as the major hypothesis for this study, the following hypotheses can be

presented for testing.

Hypothesis 1

Based on the information provided in chapter 3, it is hypothe-
sized that there will be significant distributional differences between
shipwrecked vessels of the iron, grain, and coal trades. This hypothesis
means that there would be distinguishable patterns for each of the three
industries, and that these patterns would be spatially nonrandom. In
order to test this hypothesis, it is first necessary to determine if the
patterning of wrecks is spatially random or nonrandom, and if nonrandom,
to describe the actual patterning. Chapter 7 will discuss reasons for

randomness or nonrandomness of patterning.

Test for Hypothesis 1
The procedure used to test hypothesis 1 will be to compare
frequency distributions of each industry to a Poisson distribution in

order to test for random or nonrandom patterning. To do this, the

128

locations of wrecks (both total losses and salvaged vessels) for each
industry were plotted onto the 65-cell grid and the observed frequencies
for each cell recorded. The distribution of vessels for all three
industries combined can be seen in Appendix Ca From this table, a
frequency listing of the number of cells that contain zero, one, two,
three . . . n wrecks per cell can be generated (see Table 11). The mean
can be computed by summing the multiple of the wrecks per cell (column
1) times observed frequency (f, column 2) and dividing by the number of
squares in the grid (65). In this case, the mean is 2.046. Using the
mean and it's function, the expected frequency (f) can be generated.

This frequency is distributed in a Poisson.manner and is representative

of a random distribution.

Table 11. All commodities: total and salvage combined

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 26 8.40
'1 18 17.19
2 8 17.53
3 2 11.99
4 3 6.13
5 2 2.51
6 1 .86
7 O .25
8 1 .06
9 O .02
10 0 .003
ll 0 ----
12 1 ------
13 O ------
14 3 ----

Using the Kolmogorov-Smirnov test for goodness of fit, the
cumulative observed frequency (F), cumulative expected frequency (F),
and cumulative deviation (F-F) can be computed. Table 12 provides a

synopsis of the results.

129

Table 12. Poisson results for all Lake Superior shipwrecks

E? 82 CD D

2.046 13.135 6.420 .28232

As can be seen in this table, the very high 82 and CD demonr
strate a clumped distribution not in Poisson fashion. The D value of
.28323 is sufficient for a rejection of the null hypothesis that "the
distribution of Lake Superior shipwrecks is randomly patterned" at the
.01 (.19877) level of significance. Therefore, Lake Superior ship-
wrecks clearly have a nonrandom spatial distribution. The fact that
large clusters of wrecks occur in zones 36, 54, 56, 58, and 59 (see

Appendix C), indicates nonrandom distribution.

Commodity A - Iron
The plotting of the iron industry wrecks on the Poisson grid
system resulted in the distribution shown in Appendix C-l. From these

frequencies, Table 13 generated the following results.

Table 13. Iron: total and salvage combined

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 43 22.14
1 9 23.84
2 6 12.84
3 1 4.61
4 1 1.24
5 1 .27
6 O .05
7 1 .01
8 O .001
9 1 —----
10 1 -----
11 1 ----

130

The mean, variance, coefficient of dispersion, and D statistic

were computed from this table (see Table 14).

Table 14. Iron: salvage and total losses combined

3? 32 CD D

1.077 5.517 5.123 .32090

The high 82 and CD values indicate a nonrandom distribution.
The goodness of fit test confirms this, and the D value calls for the
rejection of the null hypothesis "that the spatial distribution of iron
industry shipwrecks is randomly patterned" at the .01 level of
significance (.1988). Shipwrecks of the iron industry are therefore
highly patterned.

The clustering of ironrrelated vessels in zones 36, 54, 56, 58,
and 59, as seen in Table lc, Appendix C, is the likely cause of this
nonrandom distribution. These are the same zones that were most fre-

quently occupied by the distribution previously discussed.

Commodity B - Grain
The plotting of grain trade vessels on the Lake Superior grid
resulted in the distribution shown in Appendix C-1c. From these

frequencies, Table 15 was generated.

Table 15. Grain: total and salvage combined

 

 

 

number of observed expected
wreck/cells frequencies frequencies
0 42 39.73
1 18 19.55
2 3 4.81
3 0 .79
4 2 .10

131

Table 16 summarizes the statistical computations derived from this

table.

Table 16. Grain: salvage and total losses combined

1' 32 as D

.492 .711 1.445 .03490

The low variance and coefficient of dispersion indicate slight patternr
ing, but the D statistic illustrates that the pattern (if any) is not

significant at the .05 level (.16567). Therefore, the null hypothesis
that "grain vessels will be spatially distributed in a random (Poisson)

fashion" is accepted.

Commodity C - Coal
The distribution of coal carrying vessels on the Lake Superior
grid system resulted in the distribution shown in Appendix C - 1a.

From these, distributions Table 17 was computed.

Table 17. Coal: total and salvage combined

 

 

 

number of ' . observed expected
wrecks/cells frequencies frequencies
0 45 39.11
1 12 19.87
2 5 5.05
3 2 .03
4 0 .004
5 1 .00]

Statistical computations derived from this table are summarized as

follows:

132

Table 18. Coal: salvage and total losses combined

'; 82 CD D

.508 .896 1.764 .09060

The low variance and coefficient of dispersion indicate little patternr
ing of vessels within the grid. The D value is not significant at the
.05 level (.16567), and the null hypothesis that "coal-carrying vessels

will be randomly distributed" is accepted.

Hypothesis 1 Summary and Results

The results of the four Poisson tests indicate that the iron
industry wreck distributions are highly patterned. Five specific
locations add to this patterning. In contrast, the grainr and coal-
related vessels are random in distribution and do not exhibit the
cluster patterning found with iron. Hypothesis 1 is therefore rejected
because all commodity-related distributions are not nonrandomly
distributed. The interpretation of the results shown here will be

undertaken in the following chapter 7.

Hypothesis 2

This hypothesis states that there will be significant distri-
butional differences of shipwrecks by chronological period. It would
be expected that iron, grain, and coal vessels would distribute
differently for each period and that shipwrecks as a whole would like-
wise be distributed differently for each period. In order to confirm
this hypothesis, it would be necessary to demonstrate that wreck dis-
tributions are nonrandom for each period and that these patterns are

different for each period under consideration.

133

Test for Hypothesis 2
In order to evaluate for patterning, the Poisson test was

initiated against various shipwreck categories for three chronological
periods. Since the focus for this study is the period 1855-1930,
three equal 25-year segments were created on arbitrary grounds. These
segments were 1) 1855-1879, 2) 1880-1904, and 3) 1905-1929. For each
of these periods, shipwrecks were categorized further by separate
commodity affiliation and by combined industry totals to form.the
following 12 groupings:

1. Iron - 1855-1879

2. Iron - 1880-1904

3. Iron - 1905-1929

4. Grain - 1855-1879

5. Grain - 1880-1904

6. Grain - 1905-1929

7. Coal - 1855-1879

8. Coal - 1880-1904

9. Coal - 1905-1929

10. All vessels - 1855-1879

11. All vessels - 1880-1904

12. All vessels - 1905-1929
These groups were then plotted separately onto the 65-cell Poisson grid
for Lake Superior, and frequency distributions were recorded. These are
shown in Appendix C, 2a-1. The means and expected values can be
tabulated from these tables using the Poisson formulae previously

discussed. Tables 19-30 illustrate these distributions.

Table 19.

Table 20.

Table 21.

Table 22.

Iron 1855-1879

number of
wrecks/cell

buN—O

Iron 1880-1904

number of
wrecks/cell

 

O‘LnkwN—‘O

Iron 1905-1929

number of
wrecks/cell

134

observed
frequencies

6

t—INu—no—Io

observed
frequencies

expected
frequencies

 

 

49
11

NCO-IN

observed
frequencies

53.22
10.64
1.06
.07
.004

expected
frequencies

 

41.04
18.88
4.36
.67
.08
.007

expected
frequencies

 

 

 

\OWNO‘UI-l-‘WN—‘O

Grain 1855-1879

number of
wrecks/cell

 

0
l

5

"OOOOONU’NN

observed
frequencies

 

61
4

42.24
18.21
3.92
.56
.06
.005

expected
frgquencies

 

61.16
3.67

Table 23.

Grain 1880-1904

number of
wrecks/cell

 

Table 24.

Table 25.

Table 26.

0
l
2

Grain 1905-1929

number of
wrecks/cell

0
1
2

Coal 1855-1879

number of
wrecks/cell

0
l
2

Coal 1880-1904

number of
wrecks/cell

“NI-‘0

1'35

observed
frequencies

expected
frequencies

 

 

52
10
3

observed
frequencies

50.82
12.50
1.53

expected
frequencies

 

 

54
10
1

observed
frequencies

63
l
1

observed
frequencies

56
6
2
l

54.02
9.99
.93

expected
frequencies

,62.08
2.86
.07

expected
frequencies

53.24
10.65
1.07
.07

Table 27.

Table 28.

Table 29.

Coal 1905-1929

number of
wrecks/cell

 

0
1
2

Iron, grain, and coal combined 1855-1879

number of
wrecks/cell

 

M¢UNHC

Iron, grain, and coal combined 1880-1904

numbervof
wrecks/cell

 

\DQNGUI-l-‘UNI-‘C

136

observed
frequencies

 

50
13
2

observed
frequencies

 

5

NOI—‘NWN

observed
frequencies

expected
frequencies

50.02
13.10
1.72

expected
frequencies

 

 

41
11

HI—‘OOHOO‘L‘

47.78
14.72
2.27
.23
.02
.001

expected
frequencies

 

26.22
23.80
10.81
3.27
.74
.14
.02
.003

137

 

 

 

Table 30. Iron, grain, and coal combined 1905-1929
number of observed expected
wrecks/cell frequencies frequencies
0 36 27.04
1 15 23.72
2 10 10.32
3 2 3.02
4 1 .66
5 O .12
6 O .02
7 7 .002
8 O ------
9 O ------
10 O -----
ll 0 ------
12 1 ------

Using these expected values, the variance, CD, and D can then be
calculated using the respective formulae. A summary of the results

can be seen in Table 31.

Table 31. Summary of results
_. 2

Group Period g g_ Q 2

Iron 1 .200 .560 2.800 .10431
Iron 2 .462 1.325 2.868 .12246
Iron 3 .431 1.631 3.785 .15015
Grain 1 .060 .058 .963 .00300
Grain 2 .246 .278 1.130 .01800
Grain 3 .185 .181 .980 .00080
Coal l .046 .075 1.626 .01415
Coal 2 .200 .314 1.570 .04300
Coal 3 .262 .255 .972 .00300
All Vessels 1 .308 .982 3.190 .14185
All Vessels 2 .908 3.038 3.346 .22739
All Vessels 3 .877 2.816 3.210 .13785

The critical value for D, significant at the .05 level, is
.16567 so that only one of these 12 categories is nonrandom. The null

hypothesis that "vessels in these categories will be distributed in the

138

Poisson fashion" was therefore accept in the 11 cases. The one
exception--a11 vessels, period 2 l880-l904-—was the only case in which

the distribution formed a nonrandom pattern.

Hypothesis 2 Summary and Results

The results of the Poisson tests indicate that for the most
part, shipwreck distributions for individual chronological periods
have a random spatial patterning. Therefore, Hypothesis 2 is rejected.
In the case of category 11 (all vessels 1880-1904), a significant
pattern of vessel distribution was noted with major clustering of
vessels in zones 54, 58, and 59. Clustering was also noted for some
of the following categories, and those zones of most frequenc occurrence

are as follows:

Categogy Zone
1 58
2 58,59
3 36
10 54,58
11 54,58,59
12 36

Despite the clustering noted above, in all cases except for category 11

it was not sufficient to cause Significance at the .05 level.

Hypothesis 3
As stated in chapter 4, this hypothesis suggests that there will
be significant distributional differences between related cargo types
within the grain trade and iron industry. For example, the distribution
of grain— and flour-related vessels would be hypothesized to be differ-
ent, as would be ore- and pig iron-related vessels. When this

hypothesis was originally proposed, the extent of the data base was not

139

known.: Since that time, it has been learned that only three vessels
carrying pig iron were lost and that a small proportion of grain trade
vessels carrying flour or non-wheat cargoes were lost. Therefore, the
sample size is not sufficient to evaluate distributions using
statistical measures. A cursory examination of distributions of
commodity-related variants did not reveal any patterning that appeared
to be clustered.

Although this hypothesis is not testable given the small sample
size, it is still a valuable question to pose, and may be useful if

applied to a sample of vessels from another geographic region.

Hypothesis 4

.Hypothesis 4 states that there will be an inverse association
between the length of the nagivation season and the frequency with
which vessels of different commodities were lost. That is, iron-,
grains, and coal-carrying vessels should all be lost with greater
frequency during years with short navigation seasons than in years with
longer seasons. In order to confirm this hypothesis, it is necessary
to demonstrate that an association exists and that the association is

inverse.

Test for Hypothesis 4

 

The years between 1855 and 1930 were divided into three
groupings based on the length of the navigational season for each
respective year. Since the range for season length was 1959251 days
per year, this span was divided into three arbitrary segments of equal
length. These were:

1. Seasons with 213 days per year or less - short seasons

I40

2. Seasons with 214-233 days per year - average seasons

3. Seasons with 234 days per year or more - long seasons
All vessels were placed into these categories based on the year of their
loss and the length of season for each specific year. Table 32 lists
the calculation of season length for each year.

For each of the three groups formed, vessels were further sub-
divided into commodity association so that nine categories resulted.
The frequencies of vessels in each of these categories were then
tabulated and a contingency Chi Square test administered with the
results shown in Table 33. A cumulative X2 value of .441 was not
found to be significant at the .05 level with four degrees of freedom
(9.49). Therefore, the null hypothesis that "there will be no
association between the length of navigational season and differential
loss in each industry category" is accepted. Iron-, grain-, and coal-
associated vessels were lost in approximately the same proportions in

each of the long, average, and short navigational seasons.

Hypothesis 4 - Summary and Results

Based on the results of the X2 test, hypothesis 4 is rejected.
Rather than being an inverse association between vessel loss and length
of season, a direct association was indicated. For each industry, the
longer the period of navigation, the higher the frequency of loss.
Years with short navigational seasons experienced fewer losses of
vessels than years with long seasons. The reason behind this ess-ci-tion
may beta function of the gradual lengthening of navigational seasons in
general—-resu1ting in more years with long seasons and hence more total
wrecks. This explanation seems more likely than the assumption that a

few extra days would cause a significant increase in vessel loss.

Table 32.

Year

1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892

141

Length of navigational season

Length of
Season

163
208
205
216
209
198
195
214
214
236
217
216
240
216
204
199'
196
203
203
206
211
208
216
201
211
227
224
231
209
243
234
210
233
227
225
236

Year

1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929

Length of

Season

218
233
230
236
234
245
230
237
229
262
250
221
246
248
232
230
234
247
237
239
243
238
246
243
237
234
250
251
261
251
235
242
239
221
237
224
236

142

Table 33. Results

Iron

Grain

Coal

Iron

Grain

Coal

Iron

Grain

Coal

Short Season

 

 

 

observed expected 7X2

7 7.71 .07

4 3.60 .04

4 3.70 .02
Average Season

observed expected X2

22 22.62 .02

10 10.54 .03

12 10.85 .12
Long Season

observed expected X2

44 42.67 .04

20 19.87 .001
19 20.46 .10

143

Hypothesis 5
This hypothesis states that there will be no significant dif-
ferences in the frequency of stormrrelated shipwrecks between vessels
affiliated with each.specific commodity. That is, iron industry
vessels should be affected by storms at the same rate as the grain or

coal trade vessels.

Test for Hypothesis 5

 

The test for this hypothesis can be accomplished using a Kx2
Chi Square analysis, which tests the extent to which vessels of the
iron, grain, and coal trades are differentially associated with.loss
due to storm. Data from.Appendix B are used to determine which
vessels were lost in stormrrelated accidents, and then those vessels
are categorized by industry affiliation.

The results for the)!2 test are given in Table 34.

Table 34. Results of X2 test

Commodity Storms No Storms Total _1_’_ _P_A_
Iron 35 38 73 .4795 16.783
Grain 28 7 35 .8000 22.400
Coal 23 12 35 .6571 15.113
86 57 143 54.296

The.X2 value of 10.75 is significant at the .01 level with two degrees
of freedom (9.21). Therefore the null hypothesis that "the ratio of
vessels lost in storms to those vessels not lost in storms does not
vary by more than chance from one industry to another" is rejected.

Clearly, some industries are more affected by storm loss than others.

144

Hypothesis 5 - Summary and Results

 

Based on the results of the.X2 test, hypothesis 5 is rejected.

The key differentiations lie in the fact that grain vessels were
affected by storms 80 percent of the time as compared to 48 percent

for iron and 66 percent for coal. These figures are important because
storms may have a randomizing effect on vessel distributions by blowing

vessels off course.

Hypothesis 6
This hypothesis states that there will be significant distri—
butional differences between vessels of different ports of origin. In
order to confirm this hypothesis, it must be demonstrated that vessels
from particular points of origin (during the voyage in which they were
lost, not their point of registry) are nonrandomly distributed. If
they are patterned in their distribution, they must also be patterned

in different spatial configurations.

Test for Hypothesis 6

Since iron and grain are the only two downbound cargoes, and
coal moves upbound, only data on the former two commodities will be
used to test this hypothesis. The reporting of vessel origins was not
detailed in vessel case studies, so only 85 vessels are available for
use. Destinations for coal vessels are so poorly recorded (only
approximately 15 percent) that again, the decision was made to
eliminate coal vessels entirely.

As the first step for the testing process, the five major
ports on Lake Superior were used as origin points for lost.vesse1s.

The five ports of Marquette, Ashland, Two Harbors, Duluth/Superior, and

145

Port Arthur/Fort William accounted for 99 percent of vessel origins;
other ports were not considered. In the case of Duluth/Superior and
Port Arthur/Fort William, the ports have been combined because of their
relative proximity to one another. All the vessels from each of these
five ports were plotted onto the grid system and frequency distributions
for each can be found in Appendix C, 6a-e. From these frequencies,

Tables 35 through 39 were generated with the following results.

Table 35. Marquette

 

 

 

number of observed expected
wrecks/cell frequencies . frequencies
0 . 58 45.65
1 2 16.16
2 2 2.86
3 0 .34
4 O .03
5 1 .002
6 2 ------

Table 36. Ashland

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 59 54.90
1 2 9.28
2 3 .78
3 l .04

Table 37. Two Harbors

 

 

 

number of Observed expected
wrecks/cell frequencies frequencies
0 61 61.90

1 4 3.76

146

Table 38. Duluth/Superior

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 43 35.68
1 13 21.41
2 5 6.42
3 3 1.28
4 0 .19
5 0 .02
6 0 .002
7 1 ------

Table 39. Port Arthur

 

 

 

number of observed expected
‘wrecks/cell frequencies frequencies
0 57 57.47
1 8 7.07

From these tables, the means, variance, coefficient of dispersion, and

D statistic were computed and summarized in Table 40.

Table 40. Results

 

Port x 32 CD D

Marquette .354 1.521 4.296 .19000
Ashland .169 .326 1.930 .06307
Two Harbors .062 .058 .940 .01015
Duluth .600 1.363 2.272 .11261
Port Arthur .123 .108 .876 .00708

Hypothesis 6 - Summary and Results

It can be seen from this table that Marquette is the only port
associated with a nonrandom distribution. The high variance in relation
to the mean, high coefficient of dispersion and the significant D

statistic are illustrative of a nonPoisson distribution. Vessels from

147

Marquette are therefore nonrandomly patterned and the null hypothesis
that "vessels from Marquette will be randomly distributed across the
grid" was rejected at the .02 level of significance (.18525). For the
ports of Duluth, Port Arthur, Ashland, and Two Harbors, there were
insufficient grounds for rejecting the null hypothesis and the
patterns are therefore random.

As can be seen from Table 6a, Appendix C, the Marquette distri-
bution is clustered in zones 54, 56, and 58 and is largely responsible
for the nonrandom distribution. Although vessels from Duluth clustered
at zone 36, this alone was not sufficient to reject the null hypothesis.
The other ports exhibited no major clusters, partly because of a low
sample size.

The results of this testing allow for the rejection of hypothesis
6 since in four out of five cases, the distributions were random and
nonpatterned. There appears to be no relationship between the origin
of vessels and spatial patterning of the resulting shipwrecks. The
implication of these results may be that the Marquette--Sau1t Ste.
Marie route is more conducive to patterning of loss in some way than

the other transportation routes.

Hypothesis 7
This hypothesis postulates that there will be significant
differences between the types of vessels used in the iron trade, grain
trade, and coal trade. What is therefore expected is that different
types of vessels will be associated with different commodities or will
be used in varying frequencies. In order to confirm this hypothesis,
it will be necessary to first establish that some vessel types are

associated with particular commodity trades and then to evaluate

148

if these associations are different for each commodity and/or vessel

type 0

Test for Hypothesis 7

 

In the testing of this hypothesis, vessels were categorized by
commodity and the differential frequencies of vessel types in each
category were recorded. Five categories of vessels were recognized--
schooners, schooner-barges, wooden steamers, steel steamers, and
miscellaneous vessels. This last category was composed of a wide
assortment of vessels, including propellers, iron and composite
steamers, whalebacks, lumber hookers, etc. in order to create a
category of sufficient size to undergo statistical analysis.

The breakdown by commodity and vessel type resulted in the
creation of a contingency table with relative observed frequencies
of vessels falling into each of the cells. Analysis using a
contingency Chi Square test of association resulted in Table,41.,-.,
The Chi Square values in each cell illustrate the degree of association
between the various elements. The summation of all Chi Square values
resulted in a value of 19.21, which with eight degrees of freedom, is
significant at the .05 level (15.5). This value is sufficient to
reject the null hypothesis that "there is no association between
commodity type and vessel." An association clearly exists between these
two variables.

The X2 values within each cell can be interpreted for a finer
analysis of the association. The following relationships appear to be

the most prominent:

149

mod

am

an

mm

cm

mm

mm

 

nn.m o.“ N

no. m.~ w

00. a.m m

Ho. m.o o
NH.¢ m.m 0H

Nx venomous mo>komno

Hmoo

 

mm
mm.~ o.“ MN:
on. m.m o
mm.H m.w ma
mm.~ m.o N
oo.H m.m m

«N wouooexo vo>uomao
:Houw

 

 

mm A¢HOH

no. o.mH 0H Honuo
msooeoaaoomfiz

co. o.oH NH uoaooum
Hmmum

mm. H.¢H ma Hoammum
emeooz

om.H m.ma ma owume
Iuoeooeom
Hm. ¢.HH a Hmeoonom
Nx venomexm vo>ummno some Hommo>

eouH

endomom .He magma

150

schooners w coal

schooner—barges - grain, iron

wooden steamers — grain

#WNH

steel steamers - equal for all commodities

5. miscellaneous vessels - grain, coal

Hypothesis 7 - Summary and Results

The results of the testing confirm hypothesis 7 because an
association clearly exists between vessel types and certain commodity
groups, and because this association is different for each group. The
iron industry appears to lose proportionally higher numbers of
schooner-barges; the grain trade loses a larger number of vessel types,
including wooden steamers and miscellaneous vessels; and the coal trade
primarily loses schooners. Fewer than expected deviations occur, with
the grain trade using fewer schooner-barges and the coal trade fewer

miscellaneous vessels.

Hypothesis 8

Hypothesis 8 states that shipwrecks will reflect the changes
in vessel type that occurred over time. This proposition is very
general and is open to several possible types of testing and interpre-
tation. Implied in this statement is that relative frequencies of
vessel types will change over time and that these changes will
parallel those changes in vessel technology described in chapter 2.
The general sequence of loss should approximate the sequence of

development that is documented for the Great Lakes.

151

Test for Hypothesis 8

The test for this hypothesis will be accomplished through an
analysis of the association between various vessel types and three
periods of chronological development. Since this study is concerned
with the period 1855-1930, three chronological segments of equal size
(25 years each) were selected: period 1, 1855—1879; period 2, 1880-
1904; and period 3, 1905—1930. vessels were categorized within these
three periods and by vessel type, and relative frequencies were
tabulated from the data tables provided in Appendix B. The resulting
contingency table was the analyzed using a contingency Chi Square,
with.the results shown in Table 42. The cumulative X2 value of 81.81
is significant at the .005 level with eight degrees of freedom (22.0).
But there are several problems with the analysis. The small sample
size for period 1 (1855-1879) resulted in four expected values below
five, which have a tendency to distort the tail of the distribution.
This test also assumes that all vessel types would be represented in
each period-dwhich they were not. Steel steamers, for example, are
known to have originated during period 2, which explains the zero
observed frequency for this category during period 1. Given these
problems, it is best to simply discuss the differing observed

frequencies for each period and to de—emphasize the X2 results.

Hypothesis 8 - Summary and Results

 

Because of the unreliability of the X2 test, the interpretation
of the test results must be based on the observed frequencies. During
period 1, 13 of 21 (62 percent) vessels lost were schooners. Period 2

shows a predominance of schooner-barges (28 percent) and wood steamers

152

 

eqH No
«N Ho. mm.oH OH
Hm mm.wH mm.mH mN
He mH. no.5H oH
9N mm.H mH.HH o
NN mm.“ 5¢.¢ H
Nx pouooexo em>uomno
m ooHnom

 

 

Ho
moo. RH.oH OH
qq.m MH.mH N
mm.N um.NH «N
eN.m Ho.HH NH
aH. Nm.¢ w
NN mouoonxo eo>uomeo
N voHumm

 

HN H4909
. . mHommo>
«e on m e moomemHHoomHz
. . Hoamoum
NN q Nm e o Hmoum
. . Hoamoum
Hm n mm m H eovooz
. . mwumn
NH on m m luoeooeom
mm.¢N HN.m MH Hoeooeom
Nx mouuoexo om>uomnc mama Hommo>
H eOHuom
manmom .Ne OHQMH

153

(39 percent), while steel steamers (47 percent) dominate period 3.
These differences illustrate a chronological succession of losses
comparable to the development of vessels discussed in chapter 2. In
addition, a slight time lag is noted between the introduction of a new
vessel type and the period of loss for that same type. The vessel data
presented in Appendix B also illustrates this loss sequence. Based on
this discussion, hypothesis 8 can be accepted because vessel loss

reflects the changes in type that occurred over time.

Hypothesis 9
This hypothesis proposes that there will be significant dif—
ferences in the extent to which iron, grain, and coal were affected by
different types of loss. The four major types of loss--grounding,
foundering, fire, and collision--should therefore be differentially
associated with each of the commodities, and an association between

commodity and loss type must be demonstrated to confirm this hypothesis.

Test for Hypothesis 9

 

In the testing of this hypothesis, vessels were categorized by
commodity and loss type, and the frequencies of each were tabulated and
placed into a contingency table. Overall, groundings and founderings
were most frequent with 51 percent and 27 percent of all vessels
respectively, followed by collisions and fire at 13 percent and 5
percent respectively. Because of the relatively small numbers of
vessels in the last two categories, these were combined for analysis.

A total of 136 vessels could be placed into one of the resulting nine

cells of the table.

154

The contingency table created was then analyzed using the
contingency Chi Square test, and X2 values were obtained for each
cell in the table. The results of that analysis are given in Table 43.
The X2 values of each cell illustrate the degree of association
between the various attribute elements. As can be seen in Table 43,
there are no associations of any substance within particular cells.
The summation of X2 values results in a value of 2.48, which is far
less than the 9.49 needed for the .05 confidence level with four
degrees of freedom.

So, there are no grounds for rejecting the null hypothesis that

"there is no association between commodity type and type of loss."

Hypothesis 9 - Results and Summary

The results of the testing for hypothesis 9 demonstrate that
the iron-, grain-, and coal-related vessels were not differentially
affected by specific types of loss. Grounding, foundering, fire, and
collision had approximately the same effect on each industry.
Hypothesis 9 is therefore rejected because there was no demonstration

of significant association.

155

QMH

mN

mm

mm

 

em
RN. m.o m
mo. m.m m
0H. m.mH 0N
Nx condemns eo>uomno
Hoou

 

Hm
we. o.m a
co. n.w m
mm. e.oH mH
NN wouoonxm mo>uomno
eHouu

 

an n<eoe
ac. H.MH 0H onHm
\cofimsasoo
mo. w.mH HN weHuoeesom
«a. H.wm on weHeesouu
NM pouooexo mo>uomeo some wood
eouH

muasmmm .ms qume

156

Hypothesis 10
Hypothesis 10 states that there will be significant distribu-
tional differences between vessels of different loss categories. For
example, vessels lost by grounding should be distributed differently
from vessels lost by fire, collision, foundering-and so forth. In
order to accept this hypothesis, it must be demonstrated that vessels
of each loss category are nonrandom in their spatial distributions and

that they are distributed in different patterns from one another.

Test for Hypothesis 10

Vessels were divided into four categories based on their
condition of loss, and they were plotted onto a Poisson grid system
for Lake Superior. The number of vessels occurring in each of the
65 cells were recorded, tabulated and placed in Appendix C, 10a-d.
From these tables, the expected frequencies were calculated for
analysis, using the Kolmogorov-Smirnov test for goodness of fit.

Tables 44-47 illustrate the frequencies for each loss category.

Table 44. Grounding

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 36 20.50
1 15 23.66
2 6 13.65
3 2 5.25
4 2 1.52
5 0 .35
6 1 .07
7 1 .01
8 1 .002
9 0 _..-..
10 o -----
1 1 o -----
1 2 o ----
13 1 ----

157

Table 45. Foundering

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 47 37.35
1 9 20.69
2 4 5.73
3 3 1.06
4 1 .15
5 0 .02
6 1 .002

Table 46. Collision

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 59 51.59
1 4 11.92
2 0 1.38
3 0 .11
4 1 .006
5 0 ----
6 0 ----
7 1 -----

Table 47. Fire

 

 

 

number of .observed expected
wrecks/cell frequencies frequencies
0 59 59.29
1 6 5.45

From these tables the mean, variance, CD, and D statistic were cal-

culated within the results appearing in Table 48.

158

Table 48. Results

2

 

Loss Type '2 8 CD D

Grounding 1.154 4.930 4.272 .23846
Foundering .554 1.293 2.335 .14846
Collision .231 1.008 4.363 .11400
Fire .092 .084 .916 .00400

The high variance and CD exhibited by the grounding category point to
a nonrandom pattern. This is confirmed by the D value of .23846,
which is significant at the .01 level (.19877). The values for the
other three loss categories are not significant at the .05 level
(.16567). Therefore, the null hypothesis that "loss types will be
randomly distribute " is confirmed for foundering, collision, and fire,
but rejected for grounding. The only patterned loss category is that

of grounding .

Hypothesis 10 - Summary and Results

Based on the results of the Poisson test, hypothesis 10 is
rejected because in three of four cases, the patterns are nonrandom.
In the case of groundings (the most frequent type of loss), clustering
of distributions occurred in zones 36, 54, 56, and 58. The former
zone is particularly significant, with 13 losses within this area.
Collisions appear to cluster in zones 44 and 59, while founderings
occur most often in zone 58.

The fact that groundings are the most commonly occurring type
of loss is significant because these are also highly patterned
phenomena. The patterns of weather or local geography may make a major

contribution to vessel loss in these areas. Vessels that must pass

159

these two points on a regular basis are therefore in greater danger of

loss than vessels traveling by other routes.

Hypothesis 11
This hypothesis proposes that the frequencies of vessels in
each loss category will vary by chronological period. That is, some
loss types will associate with particular chronological periods, and
those associations will change over time. In order to confirm this
hypothesis, it would be necessary to demonstrate that an association

exists and that the association changes over time.

Test for Hypothesis 11

The chronological period encompassed by this study covers the
75 years beginning in 1855 and ending in 1930. For the purpose of
this test, this period was divided into three equal segments of 25
years each to form.the following segments-1) 1855-1879, 2) 1880-1904,
and 3) 1905-1929. Within each of these periods, the frequencies of
vessels in each loss category were tabulated and placed in the cells of
the created contingency table. The three general loss categories of
groundings, foundering, and fire/collision were used as elements of the
loss type attribute. Because of low frequencies of loss categories of
loss through fire (seven for the entire 75 year period), this category
was merged with the collision category for the purpose of testing.

The contingency table of frequencies was then analyzed using

the contingency Chi Square test with the following results (Table 49).

160

mm—

nN

cc

mm

 

 

mm mm mm H<HOH
no. _m.o_ o. _N. _n.o_ N. mN. ma.m m \aoeMmeoo
mm._ _m.e_ N. oe._ _m.s_ Na No. mm.e e manomeasoe
Na. mo.om en es._ mo.om as 0.. s... m. sebaceouo
Nx wouoomwo oo>uomno Nx mouoomwo mo>uomno NM vouoomxo eo>uomeo mmwmwmmmmm

m septum

N oomumm

_ voHuom

muHsmom .mq oHan

161

The sum.of the X2 values of each cell resulted in a total value of 6.03,
which was not significant at the .05 level with four degrees of freedom
(9.49). Therefore, the null hypothesis that "there will be no associ-
ation between chronological period and frequencies of losses in each

. loss type category" was accepted.

Hypothesis 11 - Summary and Results

 

Based on the results of the X2 test, hypothesis 11 was rejected
because no association between chronological period and loss type was
found. Within each cell of the contingency table, three associations
are of particular interest: fewer than expected groundings took place
in period 2, more than expected founderings occurred in period 2, and
fewer than expected founderings took place in period 3. None of these
three associations were significant enough to influence the overall

outcome of the test.

Hypothesis 12

This hypothesis posits that there will be an association
between certain vessel types and the corresponding types of loss.
For example, schooners may have been particularly susceptible to a
particular type of loss, such as fire or grounding. The same can be
said for each vessel type in the five general categories previously
described in hypothesis 7. In order to confirm hypothesis 12, it is
necessary to demonstrate that an association exists between vessel
type and loss type and then to discuss the specific relationships

between associated elements.

162

Test for Hypothesis 12

In testing this hypothesis, the attributes of vessel type and
loss type were broken into their individual elements as was done in the
previous hypotheses 7 and 9. This breakdown resulted in a 15-ce11
contingency table, onto which relative observed frequencies could be
tabulated. After the computation of expected frequencies using the
contingency X2 method, the analysis was undertaken on each cell. Chi
Square values were obtained and can be presented as follows in Table 50.
The summation of X2 values resulted in the overall X2 value of 27.42,
which is significant at the .005 level (23.60). This is grounds for
the rejection of the null hypothesis "that no association exists
between vessel type and loss type." Therefore, there is a strong
association between these two attributes.

When individual cells are examined for their respective X2
values, the reason for the strong association becomes apparent. The
most significant association is seen between schooner-barges and
foundering, with a X2 value of 10.76. Other associations that can
account for the strong association are between 1) schooner-barges and
grounding, 2) steel steamers and foundering, 3) schooners and

collision/fire, and 4) wooden steamers and collision/fire.

-. ..-—l.....—.... -_.4 _A __,

163

cc—

 

«N

on

an

0N

NN

 

mN
No. m.q o
no. e.m m
mm.N w.o _—
Ne._ o.< N
o_.N m.m _
Nx peacocks oo>uoono
ouHm\eonHHHoo

 

mm
mo._ N.@ a
om.N q.m a
mo. o.o_ o.
on.o_ N.N @—
ON. _.o m
NN vouooexo uo>uomno
wdHuopesom

 

 

on H<HOH
. . mHommo>
mo 0 m. «— msooemHHoomHz
seasons
on._ «.0. _N Hossm
uoaooum
mo. o.ON N. emcee:
. . omumn
«o N _ e_ m luoeoonom
nowo— Go—— @— HUQOOSUW
Nx venomous oo>uomeo some Hommo>
weHoesouu
muHammm .om mHnMH

164

Hypothesis 12 - Summary and Results

 

As previously stated, the strong association demonstrated
through the testing process confirms hypothesis 12. The frequency of
observed losses as compared to expected losses has also demonstrated
that some vessel types are particularly prone to particular types of
loss. To be specific, a large proportion of schooner-barges foundered
relative to other types of loss. Wooden steamers were also more
frequently lost through collision/fire than was expected. Other
associations were less than expected, such as with schooner-barges and
grounding, steel steamer and foundering, and schooners with collision/
fire.

The relative lack of maneuverability for schooner-barges is one
possible explanation for that association, since once a barge breaks
tow (such as in a storm), it cannot navigate successfully on its own.
The correlation of wooden steamers with fire and collision are
potentially explained by the relative susceptibility of wooden vessels
to fire. Their link to collisions remains a question at the present-

time.

Hypothesis 13
As stated, this hypothesis proposes that there will be
significant distributional differences between vessels of different
types. That is, schooners, schooner-barges, wooden steamers, steel
steamers, and miscellaneous vessels will each exhibit nonrandom pat-
terning. Additionally, if nonrandom patterns are found for each of
these vessel types, it would also be expected that these patterns will

differ from one another in their spatial distributions.

Test for Hypothesis 13

165

In order to test this hypothesis, the distributions of each

vessel type were plotted onto the 65-cell grid for Poisson testing.

Tables 13a-e in Appendix C present the frequencies obtained.

51-55 illustrate the Poisson test.

Table 51.

Schooner

number of

wrecks/cell

Table 52.

\IOUIJ-‘UON—O

Schooner-barge

number of

wrecks/cell

 

Table 53.

NVGUwa—‘C

Wooden steamer

number of

wrecks/cell

observed

frequencies

5

--ooo--w4>o~

observed

frequencies

Tables

expected
frequencies

47.76
14.71
2.27
.23
.02
.001

 

expected
frequencies

 

 

50

—‘OOOOOUI\O

observed

frequencies

 

 

UI-l-‘UJN—‘C

43
16
2

1
1
2

42.92
17.81
3.70
.51
.05
.004

expected
frequencies

 

36.80
20.94
5.96
1.13
.16
.02

166

Table 54. Steel steamer

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 47 40.95
1 11 18.92
2 S 4.37
3 1 .67
4 0 .08
5 0, .007
6 l -----

Table 55. Miscellaneous vessels

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 50 . 46.30
1 12 15.69
2 0 2.66
3 2 .30
4 1 .03

From these tables, the mean, variance, coefficient of dispersion, and

D statistic were calculated with the following results.

Table 56. Results

 

Vessel Type Ԥ 82 CD D

Schooner .308 1.033 3.353 .12677
Schooner-barge .415 1.258 3.031 .10892
Wooden steamer .569 1.199 2.108 .09538
Steel steamer .462 .626 1.355 .09307
Mlscenanews .339 .593 1.750 .05692

vessels

This table shows that the D values calculated are all well
below the value needed to reject the null hypothesis that "the distri-

bution of vessel types is random." For each vessel type, the spatial

167

distribution is random at the .05 level of significance (.16567).

Hypothesis 13 - Summary and Results

Hypothesis 13 is rejected based on the lack of nonrandom
patterning for all vessel types. Although patterns were not significant
in several cases, some clustering was noted in particular zones of the
grid. For example, schooners often occurred in zone 54, schooner-
barges in 58, wooden steamers in 58 and 59, and steel steamers in 36.

This patterning is interesting but not statistically significant.

Hypothesis 14
This hypothesis proposes that there will be a direct relation-
ship between the relative frequency of traffic vessels in each industry
and the corresponding loss of those vessels for the same industries.
That is, those industries with the greatest amount of traffic should

have the highest rates of vessel loss and vice versa.

Test for Hypothesis 14

This hypothesis is very general and can be discussed through a
simple comparison of traffic frequencies with loss frequencies. In
order to accomplish this, the chronological period 1870-1910 was
selected for scrutiny because these are the only years in which all
three industries (iron, grain, and coal) were in simultaneous operation.
Before 1870, the grain and coal industries were virtually nonexistent on
Lake Superior, while after 1910, the frequency of vessel loss diminished
to such a degree that frequencies are not comparable between industries.

For this 40 year period, the total combined tonnage of the

respective commodities was tabulated based on the figures provided by

168

Williamson (1977:212-215, 234-235, 220-221)(see Appendix A). These
commodity figures can then be interpreted in terms of relative fre-
quency of traffic, assuming that vessel transportation capacities were
the same for each industry. This assumption is not difficult to make
since iron, grain, and coal were each bulk commodities and were trans-
ported by similar means. After the commodities were tabulated, the
relative frequency of vessel loss for each correSponding industry was
also tabulated using the data provided in Appendix B. The results can

be seen in Table 57.

Table 57. Results

 

 

Commodity Tonnage Z Tonnage (traffic) Vessel Losses 2 Losses
Iron 404,391,000 71 57 51
Grain 51,252,500 9 28 25

Coal 113,962,500 20 27 24

A comparison between the percent of traffic and the percent of
losses shows that iron had nearly seven times more traffic than grain
and more than triple that of coal, but had only double the losses of
both grain and coal. 0n the other hand, grain had the least traffic
but ranked second in total vessel losses. Coal, with 20 percent of the
traffic and 24 percent of the losses, was the only commodity for which

there was a rough correlation.

Hypothesis 14 - Summary and Results
Based on the comparison presented in hypothesis 14 is rejected
since there is no direct correlation between frequency of traffic and

frequency of loss for corresponding commodity groups. The figures seem

169

to indicate that ironrrelated vessels were lost at lower rates relative
to frequency of traffic than other industries. Grain experienced high

rates of loss in proportion to the amount of cargo transported.

Hypothesis 15
Hypothesis 15 states that there will be a direct association
between the frequency of traffic in a given commodity from a given
port and the associated frequency of vessels lost from that port. That
is, if 50 percent of the iron industry traffic is from Marquette, then

50 percent of the losses would also be expected to be from Marquette.

Test for Hypothesis 15

The test for hypothesis 15 can be accomplished through the
comparison of the relative traffic from Lake Superior ports and the
associated loss from those same ports. Since the iron industry is the
best recorded for production from each port, only this commodity will
be investigated. As discussed in chapter 2, the ports of Duluth, Two
Harbors, Ashland, and Marquette served as the major foci for the ship-
:ment of iron ore. The relative traffic from each port was calculated
from the production figures of the respective areas by dividing the
total yearly production from each port by the average yearly vessel
capacity. This latter figure was obtained from sources discussed in
chapter 3 for the average vessel capacity for certain period of time.
.An.average was obtained on a yearly basis by incrementally increasing
‘vessel capacity at an average rate. The annual vessel capacities for
the period 1855-1920 are presented in Appendix C, Table 15.

The result of this traffic estimation was the conversion of

production figures into traffic frequencies. These frequencies can

170

then be compared against vessel loss counts from respective ports
tabulated from the data provided in Appendix B. Thus, the following

figure for traffic and loss can be obtained.

Table 58. Traffic and loss

 

Number of Number of Number of Passages
.2255 Passages Losses per sipgle loss
Marquette 24,663 25 987
Two Harbors 5,177 4 1,294
Ashland 13,521 11 1,229
Duluth 48,254 17 2,839

These figures indicate that for Marquette, Two Harbors, and Ashland
there was an association between increased traffic and increased loss.
In those cases, about one ship in 1,000 passages was lost. Duluth, on
the other hand, was the anomalous figure, in that although it had the
largest proportion of traffic, it had only approximately one loss per

3,000 passages-nearly one third the losses of the other three ports.

Hypothesis 15 - Summary and Results

On the basis of the comparisons, hypothesis 15 is rejected
because a uniform association does not exist between traffic frequency
and extent of loss from respective ports. The extremely small sample
available for analysis may have had an impact on these results. In
addition, there are many intervening variables that have an influence on
the vessel loss counts, making them unreliable. For example, the four
ports under consideration were not operating simultaneously. Marquette
‘was the only ore-shipping port on Lake Superior for the period 1855-1884.

In contrast, Duluth began ore shipments in 1892 when vessel technology

171

was safer and when vessel types (etc.) were different. Therefore, the
data for testing this hypothesis was neither extensive nor comparable

enough to determine a clear outcome to this traffic loss question.

Hypothesis 16
This hypothesis states that there will be no significant dif-
ferences between salvaged vessels and nonsalvaged vessels in regard to
vessel type. This means, for example, that schooners, schooner-barges,

etc. should be salvaged at the same proportional rate.

Test for Hypothesis 16

In order to test this hypothesis, each vessel type was categor-
ized into two classes--salvaged and nonsalvaged, or total loss. The
number of vessels falling into these created categories were tabulated
from Appendix B and then analyzed with a Kx2 Chi Square test. Table

59 presents these results.

Table 59. Results

 

 

 

 

Vessel Type Total Loss Salvaged Total P PA

Schooner 16 6 22 .727 11.632
Schooner-barge 21 5 26 .808 16.968
Wooden steamer 31 6 37 .838 25.978
Steel steamer 12 19 31 .387 4.644
Propeller 4 3 7 .571 2.284
Composite steamer 3 0 3 1.000 3.000
Whaleback 4 3 7 .571 2.284
Miscellanw“ 8 4 12 .667 5. 336

vessels
TOTAL 99 46 145 72.126

The Chi Square value derived from this test is 20.78, which is signifi-
cant at the .005 level (20.3) with seven degrees of freedom. This is

grounds for the rejection of the null hypothesis that "the ratio of

172

salvaged to nonsalvaged (total loss) vessels does not vary by more than

chance from one vessel type to another."

Hypothesis 16 - Summary and Results

Because of the findings of the X2 test, hypothesis 16 is
rejected; the proportion of salvaged to nonsalvaged vessels does vary
from one type to another. In particular, schooners and schooner-barges
are less frequently salvaged than wooden steamers. In contrast, steel
steamers are more frequently salvaged in comparison to other types.
One explanation for this lies with the relative cost of the vessels to
their salvage costs. Schooners and schooner-barges were cost-efficient
vessels that were regarded as expendable after a few hard years of
service. But steel steamers were very expensive vessels with a higher

salvage value and would be salvaged more frequently.

Hypothesis 17
This hypothesis posits that there will be significant differences
between salvaged and nonsalvaged vessels in regard to the port of
origin (on the last voyage) of the respective vessels. In order for.
this hypothesis to be rejected, it would have to be demonstrated that
salvage is undertaken at higher rates on vessels from.aome particular

ports as compared to others.

Test for Hypothesis 17

Hypothesis 17 is not testable because of the differential
recording of ports of origin for salvaged vessels. Only a small per-
centage of salvaged vessels had this attribute recorded, so comparisons

could not be made. Had this information been available, the test could

173

have been easily completed using a format similar to that for hypothesis
16. Despite this lack of data, the basic hypothesis posed here is still
quite relevant and may prove interesting if applied to another body of

data.

Hypothesis 18
Hypothesis 18 states that there will be no significant differences
between salvaged and nonsalvaged (total loss) vessels in regard to the
month of their loss. Vessels lost in November, for example, should be

salvaged at the same approximate rate as vessels lost in July.

Test for Hypothesis 18

Using the data from Appendix B, each vessel was categorized by
salvaged versus nonsalvaged and the month of loss. Frequency distri-
butions were tabulated and a Kx2 Chi Square analysis completed. Table

60 presents these results.

Table 60. Results

 

 

 

Month Total Loss Salvage Total P PA

January 0 0 0 0.000 0.000
February 0 0 O 0.000 0.000
March 0 O 0 0.000 0.000
April 1 1 2 .500 .500
May 6 5 11 .545 3.270
June 5 5 10 .500 2.500
July 5 2 7 .714 3.570
August 5 4 9 .556 2.780
September 22 4 26 .846 18.612
October 2? 10 3? .688 15.136
November 29 15 44 .659 19.111
December 3 0 3 1.000 3.000

174

The Chi Square value derived from this test is 8.023, which is not
significant at the .05 level with 11 degrees of freedom (19.7). Even
with the elimination of the January through March months when naviga-
tion is closed, the test is still not significant with eight degrees of
freedom (15.5) at the .05 level. The null hypothesis that "the ratio
of salvage to nonsalvage vessels does not vary by more than chance by

month of loss" is accepted.

Hypothesis 18 - Summary and Results

The acceptance of the null hypothesis results in the acceptance
of hypothesis 18. There is a slight tendency for vessels lost in
September to be salvaged at a lower rate, but this is certainly not

statistically significant.

Hypothesis 19

This hypothesis states that there will be no significant distri-
butional differences between salvaged and nonsalvaged vessels for each
commodity and for all shipwrecks lost on Lake Superior. That is, sal-
vage should be uniform for all three industries and should be spatially
distributed in the same manner for salvaged as for nonsalvaged vessels.
In addition, it would be expected that there would be no significant
difference between the distribution of all total losses (nonsalvaged).

In order to reject this hypothesis, it would be necessary to
demonstrate that all three industries have a nonrandomly patterned
distribution and that these patterns are unequal in distribution. Like-
*wise, both salvaged and nonsalvaged distributions for all three

industries combined should be nonrandom and differentially patterned.

175

Test for Hypothesis 19
In order to test this hypothesis, it was necessary to divide

vessels into six major categories.

1) Iron - total loss

2) Iron - salvaged

3) Grain - total loss

4) Grain - salvaged

5) Coal - total loss

6) Coal - salvaged
In addition, two other categories--7) all commodities total loss, and
8) all commodities salvaged-were created from the six previous group-
ings. These eight categories of vessels were then plotted onto the
Poisson grid system and differential wreck frequencies were tabulated.
These appear in Appendix C, Tables 19a-h. Using these frequencies,
expected frequencies were calculated for each. The results are as

follows.

Table 61. Iron - total loss

number of observed expected
wrecks/cell frequencies frequencies

 

 

48 32.53
22.51
7.79
1.81
.31
.04
.005

'-‘O\OCD\IO\UIJ-\WN—O
—-ooo-——-oo~w~o

~—

176

Table 62. Iron - salvaged

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 52 44.23
1 8 17.03
2 2 3.28
3 2 .42
4 0 .04
5 0 .003
6 0 ----
7 l ----

Table 63. Grain - total loss

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 47 45.61
1 15 16.15
2 1 2.86
3 2 .34

Table 64. Grain - salvaged

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 57 56.57
1 7 7.86
2 1 .55

Table 65. Coal - total loss

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 51 47.07
1 9 15.20
2 3 2.46
3 2 .26

LN

 

177

Table 66. Coal - salvaged

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 57 54.03
1 5 10.00
2 2 .93
3 1 .06

Table 67. All commodities - total loss

 

 

 

number of observed expected

wrecks/cell frequencies frequencies
0 34 16.52
1 16 22.63
2 7 15.50
3 1 7.08
4 1 2.42
5 0 .66
6 2 .15
7 2 .03
8 0 .005
9 0 .0008
10 0 -----
11 1 ----
12 0 ----
13 0
14 0 ----
15 1 ----

Table 68. All commodities - salvaged

 

 

 

number of observed expected
wrecks/cell frequencies frequencies
0 44 31.96
1 11 22.69
2 5 8.05
3 1 1.91
4 2 .34
5 O .05
6 1 .006
7 0 .001
8 1 ---

178

From these tables, the mean variance, coefficient of dispersion, and D

statistic were calculated with Table 69 providing the summary results.

Table 69. Results

 

Comodity 32 32 CD D

Ironrtotal loss .692 3.290 4.750 .23800
Ironrsalvage .385 1.129 2.932 .11954
Grainrtotal loss .354 .444 1.254 .02138
Grainrsalvage .139 .150 1.077 .00661
Coal-total loss .323 .495 1.533 .06123
Coal-salvage .185 .304 1.644 .04570
All vessels-total loss 1.137 7.128 5.230 .26892
All vessels-salvage .710 2.148 3.03 .18523

This table indicates that only three of the eight categories are signifi-
cant at the .05 level (.16567) and that there are grounds for rejecting
the null hypothesis, which states that "the distributions of vessels in
each category are randomly distributed." Category 1 (Iron - total),
category 7 (all vessels - total), and category 8 (all vessels - salvage)
are the only groups found to have distributions that are nonrandom.
Tables 19a-h, Appendix C indicate the frequency of wrecks found

in each cell. For each category, some clustering can be noted, but it
is significant only in the three categories. The zones of most fre-
quent occurrence for each of these three categories are:

1. Iron - total: 56, 58, 59

2. All vessels - total: 36, 54, 56, 57, 58, 59

3. All vessels - salvage: 36, 54

179

Hypothesis 19 - Summary and Results

Because nonrandom patterns were detected in three of the eight
categories and because salvage was not uniform across all three
industries under consideration, this hypothesis can be rejected. The
iron industry was patterned in the distribution of wrecks that were not
salvaged, while both general categories of vessels (salvage and none
salvage) were likewise patterned. The grain and coal trades are both
randomly patterned for both categories within each commodity. The
results tend to indicate that salvage was randomly undertaken in

regard to all three industries under evaluation.

Hypothesis 20
According to this hypothesis, there will be no significant
differences between salvaged and nonsalvaged vessels with regard to the
type of loss that brought about the respective shipwreck. For example,
vessels lost by grounding would be salvaged at the same rate and in the
same pattern as the vessels lost by foundering. To reject this hypo-
thesis, it must be demonstrated that particular types of loss are

associated with higher rates of salvage than others.

Test for Hypothesis 20

The four major types of loss--collision, fire, grounding and
foundering-dwere divided into categories based on whether they were
salvaged or not salvaged (total loss). The resulting eight combina-
tions were then tested by the Kx2 Chi Square method, with results as

seen in Table 70.

180

Table 70. Results

 

 

 

 

Loss Type Total Loss Salvage Total P PA

Collision 14 4 18 .778 10.892
Fire 4 3 7 .571 2.284
Grounding 39 34 73 .534 20.826
Foundering 35 3 38 .921 32.235
TOTAL 92 44 136 66.237

The Chi Square value for this test was 18.47, which is significant at
the .005 level with three degrees of freedom (12.8). Based on this
value, the null hypothesis that "the ratio of salvaged to nonsalvaged
vessels does not vary by more than chance in relation to the type of
vessel loss" was rejected. There is strong association between some

types of vessel loss and the degree to which they were salvaged.

Hypothesis 20 - Summary and Results

Based on this test, hypothesis 20 is rejected because there are
significant differences in the degree to which some loss types were
salvaged. In particular, groundings were salvaged at a much higher
rate than other loss types, while founderings were salvaged at a very
low rate. Collisions were also not salvaged as often, and those
vessels lost by fire were salvaged approximately equally. One explana-
tion for these results lies with the high degree of effort and cost
needed to salvage vessels from deep water as opposed to those lost in
shallower water. Groundings are relatively easy to salvage as opposed
to founderings and collisions, which often occur in deep water and are

thus less likely to be raised.

181

Hypothesis 21
Hypothesis 21 proposes that there will be no significant differ-
ences in regard to the frequency of salvage between vessels of the
three commodities of iron, grain, and coal. For example, vessels of the
iron industry would be expected to be salvaged at the same rates as

vessels in either the grain or coal trade.

Test for Hypothesis 21

Lake Superior shipwrecks were categorized by their respective
industry affiliations and by their status as total losses or salvaged
losses. The resulting six categories were used to tabulate frequencies
from.data provided in.Appendix B, and the following table was created

and analyzed using the Kx2 Chi Square test.

Table 71. Results

 

 

 

Commodity Total Loss ' Salvage Total P PA

Iron 50 23 73 .685 34.250
Grain 26 9 35 .743 19.318
Coal 23 12 35 .657 15.111
TOTAL 99 44 143 68.679

The Chi Square value of .803 was not found to be significant at the .05
level with two degrees of freedom (5.99). Therefore, the null
hypothesis that "the ratio of salvaged to nonsalvaged vessels does not
vary by more than chance from.ene commodity category to another" is

accepted.

182

Hypothesis 21 - Summary and Results

Based on the results of the x2 test, hypothesis 21 was accepted
because there are no differences in the frequency of salvage between
the three commodities in question. There was a slight tendency for
grain trade vessels to be salvaged at a lower rate than iron or coal
(35 percent of the time compared with 46 percent and 52 percent) but
this was not significant enough to statistically demonstrate an
association. The rapid spoilage of grain after water contamination is

a likely explanation for this tendency.

Summary of Results

 

This chapter focused on the testing of the hypotheses pre-
sented in chapter 4. Of the 21 hypotheses presented, only 19 could be
tested using the available data. Two hypotheses were untestable
because of limitations in recorded data or because of small sample
sizes not condusive to analysis. The 19 hypotheses that were tested
have provided much information on the associations between the numerous
variables under investigation. This summary section will outline the
major variables that were tested, along with the general results of the

analysis for each.

Shipwreck Location
This factor is the most important in understanding the spatial
distribution of shipwrecks across Lake Superior. Seven hypotheses
dealt directly with locational data-hypotheses 1, 2, 3, 6, 10, 13, 17,
and 19. Of these, hypotheses 3 and 17 were untestable. The remaining

hypotheses were tested and revealed that:

183

1. Shipwrecks in general are highly patterned in a nonrandom
fashionr-Hypothesis 1.

2. Only vessels frmn the iron industry are nonrandomly patterned;
coal and grain related vessels are not-Hypothesis 1.

3. Shipwrecks in general are nonrandomly patterned for the period
1880-1904, but are randomly patterned for the periods 1855-1879
and 1905-1929-Hypothesis 2.

4. Iron industry vessels from.Marquette are nonrandomly patterned,
but vessels from.Duluth, Two Harbors, and Ashland are not-
Hypothesis 6.

5. Vessels lost through grounding are nonrandomly patterned, while
vessels lost by foundering, fire, and collision are randomly
patterned--Hypothesis 10.

6. Vessels of specific types are all randomly patterned--
Hypothesis 13.

7. In general, salvage was spatially patterned in a nonrandom

fashion.

Commodity Affiliation

Vessels of the iron, grain, and coal industries were hypothe-
sized to have characteristics that set them apart from one another in
relation to other variables. These relationships were explored by
hypotheses 1, 2, 3, 4, 5, 7, 9, 14, 15, and 21. Of these, 1, 2 and 3
have been discussed as they relate to shipwreck location and will not
be further elaborated on. The results of the other hypotheses are as
follows:

1. For all three industries, there was a direct correlation between
the length of the navigational season and the frequency of loss
-Hypothesis 4.

2. Vessels affiliated with each commodity were affected differently
by storms. Grain shipwrecks experienced 80 percent loss due to
storms, coal 66 percent, and iron 48 percent-Hypothesis 7.

3. Some commodities lost higher frequencies of some vessel types
than others. Iron lost more schooner-barges, grain.more wooden

steamers and miscellaneous vessels, and coal more schooners and
miscellaneous vessels-Hypothesis 7.

184

4. All commodities were equally affected by'groundings, founderings,
fire, and collisionr-Hypothesis 9.

5. Iron industry vessels were lost at a lower rate relative to
traffic, while grain.was lost at a higher rate in proportion to
traffic than either iron or coal-Hypothesis 14.

‘6. There is no clear correlation between frequency of traffic from
a given port and corresponding losses from that port-
Hypothesis 15.

7. Vessels associated with each of the three commodities were
salvaged at approximately the same rate-Hypothesis 21.

Vessel Type
The types of vessels lost on Lake Superior are interrelated with
many other variables and constitute a major element for analysis in this
study. Six hypotheses touch on vessel type in some manner-~7, 8, 12,
13, and 16. Of these, 7 and 13 have been discussed in the previous
results as they relate to location and commodity affiliation. The
results of testing for the others are as follows.

1. The chronological sequence of vessel types lost on Lake Superior
correspond with documented historic sequences with a slight time
lag indicated--Hypothesis 8.

2. An association was found between certain vessel types and
corresponding specific types of loss. Schooner-barges were
linked to foundering, and wooden steamers to collision/fire-
Hypothesis 12.

3. Some vessel types were salvaged at a higher rate than other
types. Schooners and schooner-barges were salvaged less, and
steel steamers were salvaged at significantly higher rates-
Hypothesis 16.

Loss Type
Four specific types of vessel loss apply to Lake Superior ship-
wrecks--groundings, founderings, collision, and fire. The variable of

loss type interrelates with numerous other variables and has been

included in five hypothese-9, 10, 11, 12, and 20. While hypotheses

185

9, 10, and 12 have already been discussed in relation to previous
variables, the results of 11 and 20 are as follows.
1. Loss types did not significantly vary over time-Hypothesis 11.
2. Differences were found between salvaged and nonsalvaged vessels
. in regard to the type of losses that brought about respective
shipwrecks. Groundings were salvaged most often, while
collisions and founderings were salvaged least.
Salvaged/Nonsalvaged
The element of salvage is basic to the understanding of Type B
transforms discussed in this study. Once a vessel is deposited on the
lake bottom, a number of factors affect whether the vessel becomes part
of the archaeological record. Salvage is one cultural variable that
has a major effect on this process. Six hypotheses relate salvage to
other shipwreck variables--16, 17, 18, 19, 20, and 21. Of these, only
18 has not been discussed elsewhere in this summary. The result of
hypothesis 18 is:
1. There is no difference in salvage by the month in which vessels
were lost-Hypothesis 18.
Other Variables
A number of other variables have been discussed in relation to
the hypotheses presented in this chapter. Since all of these have been
discussed in relation to the five more prominent variables (location,
industry, vessel type, loss type, salvage), they will not be further
elaborated upon. However, the following listing of variables with

their associated hypotheses are provided as an easy cross reference to

related attributes.

186

1. Port of origin - Hypotheses 6, 17

2. Date of loss - Hypotheses 8, 11, 18

3. Frequency of traffic - Hypotheses 14, 15
4. Frequency of storms - Hypothesis 5

5. Length of navigational season - Hypothesis 4

CHAPTER VII
INTERPRETATION AND CONCLUSIONS

Interpretation of Results

The broader goal of this study is to investigate the formative
process of the archaeological record as it applies to the question of
the spatial representativeness of Lake Superior shipwrecks. Therefore,
the interpretation of the results obtained through the process of
hypothesis testing should likewise be directed toward these broader
issues. The patterning of vessels noted in the previous chapter will
be discussed as it relates to these questions, and a descriptive model
will be generated to provide a detailed explanation for the nonrandom
distributions that were found.

The testing phase of this project resulted in the definition of
three major spatial distributions of particular relevance: 1) iron
industry vessels are spatially nonrandom in distribution; 2) within
the iron industry, vessels from the port of Marquette are particularly
nonrandomly patterned; and 3) salvage was undertaken in a spatially
nonrandom.manner. Each of these results has an important bearing on
the understanding of the mechanisms of deposition and decomposition that
ultimately result in the formation of the archaeological record. The
Type A and Type B transforms that are the active ingredients of this

process will be discussed in the following sections.

187

188

Type A: Depositional Transforms

The cultural and noncultural variables that result in the
initial deposition of Great Lakes shipwrecks into the archaeological
record are dealt with in Hypotheses 1 through 15. These hypotheses
provide insights into the formative process of the Great Lakes ship-
wreck record by testing the relationship between the variables respon-
sible for vessel loss. The spatial patterns noted for the iron
industry thus have their explanation at least in part from the results
of the associations noted in chapter 6. To provide an explanation for
the nonrandom patterns of iron-related vessels and the random.patterns
of coal and grain, a brief review of the other hypotheses results is
needed. The major questions to be answered in this regard are
1) Why are iron vessels spatially patterned while coal and grain are
not? and 2) Why do iron vessels pattern along the Marquette-Sault Ste.
Marie route and not along the other routes?

The main differences between iron industry vessel distributions
as opposed to grain and coal patterns is that the former are tightly
clustered in two main areas-th *western tip of/the\Keweenaw Peninsula
between Eagle River and Copper Harbor, and the gran/of western
Whitefish Point east of Deer Park. Also, smaller clusters of wrecks
occur at Marquette, Grand Island, and Au Sable Point (see Figure 13).
Of the 75 total iron affiliated vessels lost between 1855 and 1920,
only one was located on the north shore of Lake Superior. Five other
vessels were lost on the south side of Isle Royale, four of which were
later salvaged. Along the area from.Marquette to Sault Ste. Marie, a
scatter of lost vessels accounted for a large proportion of the total

vessels lost on the lake.

189

Figure 13. Ironrrelated shipwreck distributions

 

190

 

 

mujfluj

mu..1 qu~<~m

28(1 mum
:3 (m 02.2231

 

 

3(20;
02(80
. ><¢
sun’s:—

. 923:9.

‘91

v

t 15.530

0 ' I *

5%. seen
.

 

 

191

In contrast, the pattern of grain losses was significantly
different in many respects (see Figure 14). First, seven of 35
(20 percent) vessels were lost on the north shore, while several others
were lost in open water at various points across the lake. Generally,
grain affiliated vessels have a very dispersed pattern and appear to be
scattered across a major portion of the lake, as opposed to iron
vessels, which were primarily deposited close to shore (as groundings);
that is, a higher proportion of grain vessels were lost in open water.
Approximately the same frequency of grain vessels as iron industry
vessels were lost along the southern Isle Royale coast and if grouped
collectively, they constitute the only recognizable cluster of grain
wrecks on the lake.

As Figure 15 illustrates, the pattern of coal losses is some-
where between that of iron and grain. While the pattern is dispersed
in contrast to iron, small clusters of wrecks occur in several areas.
The largest of these clusters is the area surrounding Marquette Harbor,
with five losses of 35 total occurring in this locality. There is also
a scattering of vessels between Marquette and Sault Ste. Marie, but no
major clusters of wrecks occur at Au Sable or Whitefish Points as was
the case for iron. Three coal vessels were lost along the north shore
of the lake, no vessels were lost at Isle Royale, and only one was lost
on the Keweenaw Peninsula north of the ship canal.

Obviously, each commodity displays a significantly different
pattern ranging from nonrandom/clustered for iron to random/clustered
for coal to random/dispersed for grain. The Poisson tests performed
upon these commodity distributions in Hypothesis 1 confirmed these

patterns. What remains in question is the causal factors behind these

192

Figure 14. Grainrrelated shipwreck distributions

193

 

 

w.¢<1 mum
p :(m

1:

«3.2 upD—(pm

02.22::
fl
0 ,
I 4 3530::

 

o 33.0..
an 2.20

a . >(n

.3023:—

02(234.

1:339

mung-(1
03—

 

 

194

Figure 15. Coal-related shipwreck distributions

 

 

195

 

1” mm _m in o

mu:2 w—:»<~m

 
 

25‘: E...
e :3 02.22::
a.
96+ 4 .. 230.21
Iv
a '2... . . oz<§m<
- .1 325
no t. \ a...
. a. p :30
.. 202$.
03—
0
c
09.
.1
i to 33.0..
. s oz<ao
«B
a . 2.
z :22:

 

 

196

distinctive spatial arrangements. Through the results of the hypotheses
tested in chapter 6, the interpretation of these patterns can be accom-
plished to some extent. For example, the dispersed nature of grain-
related vessels could be explained by the significantly higher per-
centage of loss due to stonms for that industry. It is very possible
that storms have a dispersing effect on distributions by blowing
vessels far off course. Travel routes requiring vessels to be far from
land would particularly be susceptible to these effects since vessels
would not have the option of running for port or attempting an inten-
tional grounding in an easily salvaged area. Because grain trade
vessels were primarily from Duluth and Port Arthur, both of these
transportation routes would require a large percentage of open water
sailing. When on open water, the rapidly rising storms of Lake
Superior can often catch vessels unaware and unprepared. Riding out
these storms often required leaving course and following the most
expedient route to safety. The high frequency of fall month storms
would exacerbate this problem because this was the peak time of travel
for grain vessels. Table 72 illustrates the relative percentages of
vessels lost in.the months of October and November, when fall storms

are at their highest frequency.

Table 72. Percentage of vessels lost in October and Novenber

Z of Vessels

 

 

 

 

Oct. & Nov. of Total Lost in Total 2 Loss

Commodity of Losses Losses Oct. & Nov. Due to Storms
Iron 18 75 24 48
Grain 20 35 57 80

Goal 13 35 37 66

197

As can be seen from this table (tabulated from Appendix B data), grain-
related vessels were lost at a much higher frequency in the fall months
of navigation than either iron or coal. Coal was second in number of
losses, possibly because of the higher fall traffic needed to stockpile
coal for winter heating. Ironerelated vessels were least affected in
these months, which may help explain their similar lower frequency of
storm losses.

If storms are actually the major cause for dispersed patterning
of Lake Superior vessels, then it would be expected that given the per-
centage of losses due to storms, iron should be least dispersed, grain
most dispersed, and coal somewhere in between. This hypothesized
distribution is identical to the types of patterning previously
described for respective commodity-related vessels. In order to test
the proposition that storms disperse vessel distributions, four Poisson
tests were performed to determine if patterning is random. Given the
respective percentages of loss by storms for each commodity group
(iron 48 percent, grain 80 percent, coal 66 percent), it would then be
expected that grain should be most dispersed, coal should be somewhat
less dispersed, and iron should be least dispersed. And if storms are
randomizing phenomena, then all three industries should be distributed
in accordance with the extent to which each are differentially affected
by storms. Appendix C contains the frequency counts for these Poisson
tests. The observed and expected frequencies tabulated from the

frequency counts are as follows.

Table 73. Iron - storms

number of
wrecks/cell

\DQVGUIDUJN—O

Table 74. Grain-storms

number of
wrecks/cell

 

#UNh-O

Table 75. Coal - storms

number of
wrecks/cell

 

LnJ-‘UON—‘O

198

observed
frequencies

4

c—CDOOOONO‘NO

observed
frequencies

expected
frequencies

 

 

47
14
3
0
1

observed
frequencies

38.50
20.20
5.30
.92
.12
.01

expected
frequencies

 

47.05
15.20
2.46
.26
.02

expected
frequencies

 

 

50
10
4

0
O
1

48.54
14.17
2.07
.20
.02

199

Table 76. All commodities - storms

 

 

 

number of observed expected
wrecks/cell frequencies frequencies

0 34 18.10

1 13 23.10

2 6 14.80

3 7 6.30

4 1 2.00

5 1 .52

6 O .11

7 1 .02

8 O ---

9 0 —---

10 1 ---

11 1 -m—-

A summary of the goodness of fit, variance, and coefficient of disper-

sion for these four tests is shown in Table 77.

Table 77. Summary

2

 

Category 'E 3 CD D
Iron .523 1.727 3.302 .1615
Grain .323 .509 1.576 .0269
Coal .292 .663 2.271 .0417
All vessels 1.280 4.690 3.670 .2446

At the .05 confidence level (.1657), only the combined distribution for
all vessels is patterned nonrandomly. Each individual commodity is
therefore randomly patterned as would be expected if storms exhibited

a randomizing effect on vessel distribution. The figures in Table 77
also illustrate that ironrrelated vessels are least randomly affected

by storms, while grainrrelated vessels are most dispersed in their
pattern but not to the same extent. There is direct association between

ironrcarrying vessels and a more clustered but random distribution, and

200

grainrrelated vessels and a very dispersed distribution. Given that

80 percent of all grain shipwrecks are storm related, it is not sur-
prising that overall grain vessel distributions are random. In contrast,
iron industry related vessels are least affected by storms and exhibit
the most clustering in stormrrelated accidents. A major clustering of
these vessels was encountered for zone 58 (Whitefish Point), which
indicates a higher frequency of loss in this area for iron related
wrecks.

Of the four Poisson tests conducted, the combined category
resulted in a nonrandom pattern. This shows that the overall patterning
of vessels in stormrrelated accidents are nonrandom, with clustering
occurring in zones 35, 36, 54, and 58. The entire western edge of the
Keweenaw Peninsula, the Marquette area, and Whitefish Point were col-
lecting areas for vessels lost in storms. These results imply that
since most storms come from the west and northwest, the highest con-
centrations of lost vessels will be in areas on the windward side of
the storm path. Conversely, few vessels were lost on the leeward side
of land because they would have been in protected waters. Whitefish
Bay west of Sault Ste. Marie is an example of this. As hypothesis 1
has shown, the highest losses on Lake Superior (15) occurred in zone
59--Whitefish Bay. Of these losses, only three (20 percent) were a
result of storm conditions. In contrast, the windward face of
Whitefish Point (zone 58) had a similar 15 total losses but 11 of these
(73 percent) were caused by storm activity. Clearly, those routes
following the most windward stretches of water run the highest risk of

loss due to storms.

201

The randomizing effects of sfbrms‘may also have had a greater
impact on some vessel types than on others. Generally speaking, the
more sizable and modern the vessel, the better the chance it would have
of either avoiding storms (because of modern weather communication) or
riding out the storms while maintaining vessel course. For example,
modern ore boats regularly faced fall storms with fatalities such as
the Edmund Fitzgerald, an extreme rarity. As vessel size gradually
increased over time, the ability of vessels to weather the storms of
Lake Superior also increased. It is not surprising therefore that
hypothesis 7 illustrates a much higher frequency of loss for smaller
vessel types-schooners and schooner-barges--than for larger types such
as steel steamers. Hypothesis 8 reveals a similar trend; for each
chronological period under consideration, particular types of vessels
were lost in varying proportions. The period before 1905, when both
sailing vessels and larger steam powered vessels were in competition
on the lake, also was the time when significantly higher numbers of
sailing craft were lost. Schooners and schooner-barges were lost at
much higher rates than.the more modern and costly steel steamers. This
cost factor is especially important to an understanding of the storm
randomization process.

As discussed in chapter 3, the iron, grain, and coal trades
varied substantially in capital intensity. Iron in particular was the
most significant industry on Lake Superior with total transportation
for 1870-1910 being four times that of coal and eight times that of
grain for the same period, as pointed out in hypothesis 14. Initial
investment into the iron industry by wealthy eastern entrepreneurs and

later investment by large national conglomerates such as US Steel helped

202

to build the industry from a small pick and shovel operation in 1850 to
the premier example of capitalistic endeavor by 1900. Throughout this
process, the availability of tremendous amounts of capital for expansion
and improvements has been a significant factor in rapid development.

As pointed out by several authors, (Hatcher 1950), (Barry 1974), (Evans
1942), this financial capacity enabled the iron industry to create and
maintain the largest and most modern fleet of vessels on the Great Lakes
for the period under study. As new innovations in vessel technology
were discovered, the iron industry soon adapted them to their fleets.

As a newer and larger vessel was designed, one of the major iron
companies soon purchased it for use in the Lake Superior ore trade.
Technological innovation followed the lines of highest capital intensity
with the wealthiest companies and industries receiving modern improve-
ments before others of lesser financial ability. This process allowed
the iron industry to maintain a large modern fleet, while smaller
industries relied more upon slightly older and less modern.vessels.

When steel and iron steamers first came into use in the 18708 and 18803,
it was the iron industry that first adopted than (Barry 1974:145-188).
This was true for the majority of new bulk carriers that were introduced
throughout the period. As iron fleets were updated, older vessels were
scrapped and sold when their economic efficiency dipped below desired
standards.

Compared to the modern and wealthy iron industry, the tranSpor-
tation of grain was a poor competitor. The grain trade per se was not
an industry in the strict sense of the word. Rather, it was a process
by which grain was centralized in certain ports for shipment to other

regions. Grain was moved under contract between the purchasers of the

203

cargo and the independent vessel owners or shipping companies. Although
the grain trade was by no means simple in technology or transportation,
it was not as centralized as the iron industry, hence technological
improvements came more slowly. The decentralized nature of the grain
trade influenced its ability to make uniform capital improvements in the
fleets and as a result, grain fleets on the whole lagged behind iron
fleets in vessel size and modernization. Smaller, less modern.vessels
would, therefore, be more susceptible to loss by storms, hence the
higher loss rate.

The movement of coal was also not a true industry but a
commodity marketing system similar in many respects to that of grain.
The main difference between coal and the other two commodities is that
it moved upbound as a return cargo for both ore and grain vessels. The
iron industry often had close ties with coal producing areas, and ore
fleets commonly carried coal on the upbound leg of the journey. The
iron industry as a whole used large amounts of coal for power in the
mining process, so large quantities were required at the same ports
from which ore was shipped. In other cases, coal was transported by
independent vessel owners to minor ports for use in home heating and
nonindustrial power. The combination of links to the iron industry as
well as links to smaller transportation operations had the cumulative
effect of placing the coal trade somewhere in between iron and grain in
overall vessel fleet modernization. Improvements in iron fleets
generally resulted in improving the movement of coal, but not always.

The resulting distribution of coal-related vessels can be
viewed as a by-product of a vessel fleet that is only partially

susceptible to the randomizing effects of storms. Iron-related coal

204

vessels would theoretically withstand storms better than independent
vessels. It could be suggested that more clustered coal vessels (i.e.,
between Marquette and Sault Ste. Marie) should be those having iron
industry links, while vessels in a more random.pattern should be inde-
pendently owned or at least nonironrrelated vessels. Unfortunately,
available data is insufficient to test this proposition at the present
time.

Other information relating to this question results from the
test of hypothesis 2, which found that vessels for the period 1880-1904
were spatially patterned in a nonrandom distribution, while vessels for
the years 1855-1879 and 1905-1929 were randomly distributed. If
capital intensity and spatial patterning are closely related as sug-
gested in this model, then it is possible that chronologically, the
periods of largest capital inputs will be the periods of most significant
distributional agglomeration. The extremely small sample sizes that
result when each of the three industries are subdivided into three
chronological periods may have an effect on the Poisson tests for dis-
tributions. As should be recalled from hypothesis 2, none of the three
industries demonstrated nonrandom patterning for any of the periods
under consideration. Although not significant at the .05 level, the
iron industry had consistently higher D values, indicating more
patterning than either grain or coal. When all commodities are come
bined, the sample size is much larger and the results more reliable.
With this combined sample, the three chronological periods indicate
some degree of patterning with the middle period 1880-1904 being
significantly nonrandom beyond the .005 level of confidence. This

period is interesting since it corresponds with the time of highest

205

capital improvements of the bulk carrier fleets. In 1882, the 92253
was produced as the first large iron bulk carrier on the lakes (Barry
1974). Between that time and 1905, the shipping industry witnessed an
unprecedented period of expansion and by the beginning of the twentieth
century, the 600-foot carrier had become standard. As discussed in
chapter 3, the economic consolidation of iron companies after the 1893
depression spurred this period of capital improvement of fleets. The
iron industry led this expansion period and innovations began to filter
slowly into other industries. If capital investment is linked with
vessel patterning, it would then be expected that vessels for the
period 1880-1904 would be more patterned than for either of the previous
or later periods.

The scenario thus presented as an explanation for differential
vessel distributions by commodity type can be summarized by the proposi-
tion that there is a direct correlation between the capital intensity
of an industry and the degree of spatial patterning that will occur. A
stronger financial base results in more modern vessels less susceptible
to the randomizing effects of weather. A counterpart to this model is
that the route of transportation is likewise directly related to this
loss process because it also has a bearing on the extent to which storms
will affect vessels.

Hypothesis 6 illustrated that of the four major iron routes,
only vessels from Marquette exhibited nonrandom patterning. This implies
that some routes are more conducive to loss than others. The reason or
reasons behind this are not revealed in the hypotheses tested in this
study, although the results of Hypotheses 1, 6, and 10 consistently

point to clustering of lost vessels in zones 36, 54, 57, 58, and 59.

206

These five zones encompass the areas of western Keweenaw Point,
Marquette, and Whitefish Point. A brief analysis of the 64 vessels
found in these five zones reveal the following frequencies of losses

for each industry.

Table 78. Frequency of losses by industry in five zones

 

 

Losses in 2 Total Industry Losses in Zones
CommodiEy 5 Zones Losses Due to Storms
Iron 41 59 16-39
Grain 12 38 9-75
Coal 11 33 10-91

This table illustrates that approximately 45 percent of all Lake
Superior shipwrecks occurred in these five zones. When subdivided by
commodity affiliation, it can be seen that 59 percent of all iron
vessel losses occurred in these zones compared to 38 percent for grain
and 33 percent for coal. The extremely high percentage of losses for
iron is further illustration of the highly patterned nature of ironr
related wrecks as opposed to either grain or coal. Of special interest
is the number of losses due to storms for vessels in these zones.
Whereas total iron losses were highest in these zones, the percentage
of vessels lost by storms was the lowest at 39 percent. The 75 percent
for grain and 91 percent for coal indicate a significantly higher pro-
portion of losses in these same areas. As previously mentioned, this
points toward the lower rate of loss due to storms within the iron
industry. This lower storm loss rate, along with the clustering effect
noted in these five areas, also supports the proposition that patterning
would be more nonrandom for iron because of the relative lack of
patternrrandomizing storms. Since a lower percentage of iron vessels

were lost because of storms in these five zones, the patterning of iron

207

vessels will likewise be more consistent with the routes of transporta-
tion for that industry.

The iron industry focused on ore shipment from four major ports:
Duluth, Two Harbors, Ashland, and Marquette. The transportation routes
for ore shipment are such that the Two Harbors, Ashland, and Duluth
routes intersect off shore of Eagle River on the Keweenaw Peninsula.
Before this intersection, the majority of routes are across open water
except for the western Apostle Island groupings. After the inter-
section of traffic has taken place, all vessels from these three ports
make several major course changes after rounding the tip of the
Peninsula in order to make the southern course for the Sault. One
other course change is made off Crisp Point on.Whitefish Point before
directing vessels to locks and the St. Mary's River. The Marquette
route is the only Lake Superior route that follows the southern coast-
line in close proximity to shore for the entire length of the route.
After a relatively minor course change at Au Sable, no other changes
are made until Crisp Point when all traffic on Lake Superior merges.

Zones 36, 57, 58, and 59 have several common features that are
unique to these areas alone and that may play an important role in
understanding the clustering of wrecks in these zones. First, both the
Keweenaw and Whitefish areas represent points where several routes
intersect and where traffic becomes more congested. In good weather,
this would present only minor navigational problems, but could prove
hazardous in conditions of poor visibility. Similarly, both areas
require major course changes that compound navigation problems in less
than perfect weather. The accompanying narrowing of navigation,

especially off Whitefish Point, also may play an important factor in

208

vessel loss. Mansfield notes that "the principal danger to navigation
on the lakes is lack of sea room, which leads vessels to run for
shelter in a storm and to seek entrance into artificial harbors, which
they are liable to miss and to strike upon the piers at the entrance to
them" (1899:378). Finally, both areas lie along windward coastlines
that take the full brunt of western storms. Being close to land at
these two points also means that the likelihood of fog is greatly
increased over more open water routes.

Although fog was a variable not directly considered in this
study, it was an element mentioned by Wolff (1979) in numerous case
studies of individual vessel losses. A reconsideration of this variable
has shown that 16 vessel losses occurring between 1855-1930 were
influenced to some degree by fog. Of these cases, 81 percent occurred
in the five zones under consideration, and all were associated with
iron industry vessels. According to the data provided in.Appendix B,

44 percent of all collisions on.Lake Superior took place within these
same five areas. Therefore, fog appears to be an.element that affects
the iron industry much more than the grain or coal industries.

A probable explanation for the high instances of fog-related
losses in the iron industry is that the routes of ore transportation
are for the most part closer to land than those of either grain or coal.
The Marquette-Sault Ste. Marie route is particularly susceptible because
the majority of it is within fog range. In contrast, the Two Harbors,
Ashland, Duluth, and Port Arthur routes are close to land only at
restricted areas such as the Keweenaw and Whitefish Points. The
clustering of wrecks in zones 36, 54, 57, 58, and 59 and the patterning
of vessels from Marquette are probablythe result of a combination of

factors.

209

Besides being in a high probability fog zone, the Marquette
route is also the only route where loss by grounding can occur in close
proximity to the actual route being traveled. As hypothesis 10
illustrated, these same zones exhibit a highly clustered pattern of
vessels in this loss type. This is as would be expected if narrowed
navigation, fog, and course changes played significant roles in vessel
loss. Therefore, the clustering of vessels in these zones and along the
Marquette route can best be explained by the combination of specific
geographic and climatic variables with cultural variables such as
navigation routes. The general pattern that emerges is that vessel
loss is highest along 1) points of course changes; 2) narrowed
navigation; 3) traffic intersections; 4) close proximity to land where
fog, collision, and groundings can be major factors; and 5) windward
eXposed coastlines.

The pattern of vessel loss presented here is Specifically
applicable to Lake Superior, but the question remains as to its pre-
dictive value in other areas. The information generated by this study
is useful in determining only general patterning of loss, not specific
geographic areas where loss takes place. In other areas, Lake Michigan
for example, the pattern of loss could be quite different because of
geography, navigation, weather, and other variables that are different
from the Lake Superior region. On the other hand, the principal agents
in determining loss patterning may be the same for all the Great Lakes.
If so, then given the variables of deposition that have been derived,
it could be possible to outline likely areas of loss for other water-
bodies. This is a proposition that can be tested during the course of

future historical and archaeological inquiry. When the concept of

210

capital intensity and patterning is introduced into this system, very
specific hypotheses can be tested for other regions and a theoretical
foundation for shipwreck studies can be laid down. Using the informa-
tion provided in chapter 3 on such characteristics as the chronological
sequence of ore types (hard, soft, Bessemer, graded, etc.) and grain
varieties (hard, soft, white, red), extremely detailed hypotheses

could be generated for testing with a data base more extensive than for
Lake Superior. With only 146 vessels available in this study, more
specific hypotheses could not be sufficiently prepared and analyzed.
However, with a sample.size of 300 vessels, models could be developed
that could be tested with historical and archaeological data.
Additionally, the information from chapter 3 can be used to date
vessels of unknown origin using cargo samples if archaeological investi-
gations on shipwrecks do occur. Therefore, the model of deposition
presented in this text is to be regarded as a general foundation upon

which more complex distributional models can be built.

Type B: Archaeological Transfonms

Hypotheses 16-21 concern the effects of salvage on Lake Superior
shipwrecks. After shipwrecks are deposited $2 2252 on the bottomlands,
a number of factors influence the degree to which they are preserved
and/or decomposed. A number of these factors such as ice damage,
decomposition due to the limnological conditions, etc. are currently
unknown and can only be investigated through direct observation of the
archaeological record. But other variables, such as salvage, are
historically documented to some extent and can be investigated within
the scope of this study. When viewed as an archaeological transfomm,

salvage is an extremely important variable because it directly affects

211

the extent of the archaeological record.

Since this study focuses upon the question of spatial repre-
sentativeness of Lake Superior shipwrecks, the element of salvage must
also be reviewed for its effect upon the distribution of lost vessels.
The major question here is to what extent salvage alters the spatial
arrangement of shipwrecks from initial pattern of deposition.

Hypothesis 19 is particularly relevant in answering this question
because it has shown that collectively, salvage is a patterned nonrandom
phenomena. Four zones in particular had heavy clusters of salvaged
vessels-zones 44, 54, 36, 59. The first two zones are port areas for
Duluth and Marquette and represent the two busiest ports on Lake
Superior. Apparently, salvage is undertaken at higher rates within
close proximity to major ports. This is not unusual since distressed
vessels can be reached soon after loss, which aids the salvage procedure.
Likewise, the relative closeness of other vessels such as tugs allows
for towing of unnavigable wrecks and the freeing of vessels from
grounded locations.

The two other locations of highest salvage frequency were the
Keweenaw Peninsula between Eagle River and Copper Harbor and the section
encompassing Whitefish Point and the waterway to the Sault. These
areas (zones 36 and 59) are also in close proximity to major ports and
salvage facilities at Houghton and Sault Ste. Marie, and both areas are
accessed within 25 miles or less of the respective ports and are
therefore quite close in relation to the vast area of Lake Superior.

The other major ports of Two Harbors, Ashland, and Port Arthur also
have several instances of salvage close to harbor facilities, although
no significant clusters of totally lost or salvaged vessels occur at

these areas.

212

Another factor that may influence the spatial distribution of
salvaged vessels is the type of loss that brought about the initial
wreck. Hypothesis 20 demonstrated an association between loss type and
extent of salvage, with groundings being salvaged more frequently than
other loss types and founderings less frequently. Remembering that
hypothesis 10 illustrated that groundings cluster in zones 36, 54, 56,
and 58, it is not surprising that salvage of grounded vessels is also
more frequent in these areas. That vessels in these zones grounded at
a higher rate than in other areas may also be an explanation for why
salvage was also conducted at higher rates in zones 36 and 54. Although
groundings are of varying degrees of severity, this type of loss is
generally more conducive to salvage because of shallower water depths.
Many groundings are minor matters for salvage and can be conducted in a
few days by simply releasing the vessel from a reef or shore and
patching damaged hulls. The tendency for founderings to be salvaged at
a lower rate is because most occur in the open, deep waters of Lake
Superior. Until the middle of the twentieth century, salvage of deep-
water wrecks was a rarity and would not have been attempted on the lake
for the time period dealt with in this study. From Appendix B, it can
be seen that only four vessels lost before 1885 were ever salvaged. In
two of these cases, Queen of the Lake and Bermuda, salvage was not
undertaken until many years after the loss had occurred (Wolff 1980:
54-55). Salvage on Lake Superior was therefore not a factor before the
18803, so spatial patterns for the early period of the lake would
remain unaffected by this variable.

The variable of vessel type was shown in hypothesis 16 to be

another factor that was differentially affected by salvage. While some

213

vessel types such as schooners and schooner-barges were salvaged at low
rates, other types such as steel steamers were salvaged considerably
more often. The results of this hypothesis illustrate that there is a
strong tendency for more valuable vessels to be salvaged at a higher
rate. For example, steel steamers are larger, more costly to produce,
and have a higher scrap value for engines, metal and so forth. Their
larger capacity in contrast to other vessel types also means that more
cargo will be carried on a given voyage with a higher total value.
Schooners and schooner-barges were less expensive to produce, had little
"scrap" value since they were built of wood, and carried smaller amounts
of cargo in a single trip. They also had shorter life expectancies and
given the fact that most schooner-barges were conversions of older
schooners, the life span of these sailing vessels was probably reached
while in service. As hypothesis 12 has shown, schooner-barges were
particularly susceptible to loss by foundering, which may have been
accentuated by vessel age. Together, these variables make it expectable
that some vessel types were salvaged at higher or lower rates than
others. An analogy would be that most people today do not expend large
sums of money on the repair of tenryear-old automobiles that have been
in accidents. It is not cost efficient to save a vehicle that has
already reached its life expectancy, just as it is not feasible to
salvage a vessel of low value. In terms of vessel type, it is important
that the vessels that form the shipwreck archaeological record are not
reflections of the same types of vessels that were originally lost.
Taken collectively, the salvage variable is a major influence on the
formation of the archaeological record. The spatial distributions of

vessels is altered by the patterns of salvage since it occurs more

214

often near coastline areas and close to major ports. Also, some
vessel types and loss types are salvaged more often than others, which
influences the relative frequencies of vessels in the archaeological
context. As a Type B transform, the element of salvage is a signifi-

cant factor in the patterning of Lake Superior vessels.

Conclusion

While the primary goal of this study was the investigation of
spatial patterning and shipwreck formation, a broader underlying goal
was the demonstration of the applicability of shipwreck data to
nonparticularistic research. The current state of shipwreck studies
nationwide is that they are primarily oriented toward history and
architecture and not toward more holistic issues. Although this is a
slight improvement on the antiquated philosophy of the past, the
historical approach still does not address shipwrecks as reflections of
a cultural phenomena. In other regions of the country, the focus is
on chronologically earlier vessels about which little or nothing is
known in regard to architecture, artifacts, or historical movements of
people and supplies. In these areas, archaeology as history contributes
to the body of historic knowledge in a substantial manner-albeit limited
in social interpretation. Research can therefore be justified to some
extent because of the contribution to such disciplines as marine history
and architecture. In the Great Lakes region, the situation is very
different from coastal areas. On Lake Superior, for example, vessels
are primarily from.the post-1855 period, which is documented more
extensively than other regions. For many vessels, historians know the

architecture through written documents such as blueprints. Likewise,

215

the artifacts found on vessels, especially from the twentieth century,
are not particularly valuable or unique as objects of interest. For
these reasons, the shipwrecks of the Great Lakes, particularly Lake
Superior, are of little value in the strict historic sense because they
do not contribute knowledge about specific people, events, or objects

of the past. This in part is a reason for the lack of research on

Great Lakes shipwrecks by historically-trained underwater archaeologists.

While Lake Superior shipwrecks are of limited interest to
traditional historians, they have great potential for anthropologists as
reflections of culture process. The major finding of this study is
actually that shipwrecks can be used as a data base for solving
broader social questions about the past. This research has analyzed
historical case studies and applied them to propositions that involve
the past as an interrelated system. A systems approach recognizes that
the past is more than just the "big events" that historians portray.

In addition, this study has addressed several questions that
are immediately relevant to the study of Great Lakes shipwrecks from an
archaeological perspective. As a region, Lake Superior is extremely
well-researched, the documented wreck sites within its boundaries
constitute a major portion of the vessel population lost during the
period 1855-1930. Before 1855, settlement and development of the
Lake Superior region was extremely sparse and limited. The effect of
the Industrial Revolution on the area created an aspect of culture
change that is unique in comparison to other portions of the country.
Lake Superior represents a tightly confined geographic area that was
developed solely in response to industrial expansion and the need for

raw materials. The spatial configuration of towns, industrial camps

216

(iron, charcoal, lumber, etc.), and shipwrecks was a result of indus-
trialization that no other part of the country experienced. Through
the spatial analysis of this region, industrialization can be given
significance as a cultural phenomenon, rather than as the technological
phenomenon recognized by economic historians. Shipwrecks as an
industrial phenomenon therefore have great potential for providing
information on the process of the culture change that swept the Great
Lakes region.

Analysis of this shipwreck population suggests that there is a
direct correlation between the capital intensity of an industry (i.e.,
ability to make capital improvements) and the degree of spatial
patterning that results in wreck deposition. This is a hypothesis that
can be tested for the other Great Lakes and may have relevance to many
other shipwreck distributions. It is also a pr0position that may have
some use in dealing with industrial archaeological sites on land
because theoretically, technological innovations are diffused geograph-
ically across lines of capital investment. For land sites, the actual
mechanisms patterning would be very different because the randomizing
effects of storms would not apply and specific models of diffusion and
deposition would necessarily have to be generated in order to apply to
these sites. However, this proposition can be tested and may shed some
light on the mechanisms of change and spatial patterning sought by both
anthropologists and geographers.

As reflections of past cultural systems, shipwrecks can provide
a great deal of information useful to archaeological research. But this
study has also demonstrated that in many ways, shipwrecks are not

representative of every aspect of economic development and change that

217

has taken place. There are many questions that shipwreck data cannot
answer in regard to the past. Shipwrecks are a culturally structured
phenomena as are all archaeological sites, but this alone does not
assure that the archaeological record is representative in every way of
the culture from which it originated. The spatial analysis of Lake
Superior shipwrecks resulted in the rejection of many hypotheses from
the expected patterns. Vessels were differentially distributed by
industry, port of origin, vessel type, and loss type. Therefore, the
shipwrecks that now comprise the archaeological record are patterned

in specific ways that do not directly reflect all aspects of culture
change and development. Future research must consider these limitations
and ask questions that are testable, given the condition of the
archaeological record.

By expanding the transform concept into two separate processes,
this study has introduced an.emended version of Schiffer's (1972) model
of the formative process of the archaeological record. The processes
of deposition and decomposition are useful to the understanding of
shipwreck distributions. These same concepts have utility for all
archaeological phenomena since the material culture initially
deposited iu.gi£u undergoes numerous changes before the archaeological
investigation. This transformation process has a great deal of
relevance for this study because spatial patterning reflected the
cultural developments that affected the Lake Superior region.

The real importance of patterning lies in an understanding of
the interrelatedness of cultural and environmental processes that
created the present distribution of shipwrecks. The interactions of

trade routes, industries, developmental chronology, weather, land forms,

218

and a whole array of othentcultural and natural factors have produced
the spatially patterned archaeological record that archaeologists must
now study.

The implication of this study to future shipwreck research is
that a regional view must be maintained in order to fully understand
the processes of change and development that effected the entire area.
If only one small geographic area within the region had been selected
for study, an extremely biased set of results would have been obtained.
Since clustering of wreck sites is a patterning phenomenon influenced
by particular variables, the study of such a single cluster would like-
wise be representative of atypical cultural and environmental conditions.
For example, because iron industry vessels tended to cluster while coal
and grain vessels dispersed, the major concentrations of wrecks tend to
be heavily biased toward iron. Additionally, because the iron
industry lost particular vessel types at higher rates than others and
by particular types of loss, the concentrations of vessels would also
be nonrepresentative of the types of vessels that actually sailed the
region. Extreme caution.must be maintained if individual wreck con?
centrations are evaluated and interpreted outside of a regional format.
If a representative cross-section of industries and types is desired,
then spatially diverse individual wreck sites must be sought. In a
similar vein, if protection of a cross-section of Great Lakes vessels
is desired, then the protection of only major wreck concentrations will
not fulfill this desire.

Hopefully, this study has provided some useful explanations and
insights into the phenomena of Great Lakes shipwrecks. As with any

research, many questions are raised--and this study is no exception.

219

The archaeological resource that is least known and understood is that
of the shipwreck. The estimated 6,000 wrecks that have occurred in the
Great Lakes encapsulate the culture of the region, if not the nation.
This study has focused on 146 vessels-a small number in-comparison to
what may lie beneath the waters. The old adage "out of sight, out of
mind" must not apply to this phenomenon. The propositions posed and
discussed in this study are only the beginning steps toward creating a
basis for scientific research and analysis. If not, then this study

has not fulfilled its goal of demonstrating the relevance and importance

of shipwreck study for the whole of archaeological understanding.

APPENDIX A

Table 1.

Year

1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892

Sources:

APPENDIX.A

Opening and closing of navigation

Opened

June
May
May
April
May
May
May
April
April

Mansfield 1899:508; Hickok 1938:321.

13
‘4

9
18

28
7
21
2
23
6
25
1
7
15
20
27
15

Closed

Nov.
Nov.
Nov.
Nov.
Nov.
Nov.
Nov.
Nov.
Nov.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Nov.
Nov.
Nov.
Dec.
Dec.
Nov.
Nov.
Dec.
Dec.
Nov.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.

23
28
30
20
28
26
14
27
28
24
3
3
3
3
29
l
29
26
18
2
2
26
30
3
3
15
5
3
11
10
2
24

20
4
4
3
8
7

220

Year

1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930

Opened

May

April
April
April
April
April
May

April
April
April
April
May

April
April
April
April
April
April
April
April
April
April
April
April
April
April
April
April
April
April

April
April
April
April

April
April

5
17
25
16
22
13

19
26

14
13
23
27
26
12
22
24
19
23
18
20
24
24
10
19

17

19
13
29
13

14
22

Closed

Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dec.
Dan.

7

9
11

8
12
14
18
12
11
20
15
13
16
17
11
13
16
15
16
19
18
17
20
19
17
14
16
26
24
24
26
19
16
14
14
14
14

0'
I.)

Table 2. Upper Peninsula forges and furnaces

Name

 

Carp River Forge
Marquette Iron Co.

Forest Iron Co.

Collins Forge

Collins Experimental Furnace
Pioneer No. 1 Stack
Collins Blast Furnace
Northern Furnace

Pioneer No. 2 Stack
Bancroft Furnace

Morgan Furnace

Greenwood Furnace
Michigan Furnace
Schoolcraft Furnace (Munising)
Jackson No. 1 Stack
Champion Furnace

Deer Lake No. 1 Stack
Jackson No. 2 Stack

Bay Furnace No. 1 Stack
Marquette & Pacific

Peat Furnace (Excelsior)
Grace Furnace

Bay Furnace No. 2 Stack
Escanaba Furnace (Cascade)
Menominee Furnace

Carp River Furnace

Deer Lake No. 2 Stack
Cliffs Furnace

Martel Furnace

Vulcan Furnace

Iron River Furnace (Gogebic)
Weston Furnace

Gladstone Furnace

221

Dates of Operation

 

1848-1857
1850-1853
1855-1862
1855-1858
1858-1858
1858-1892
1858-1873
1860-1910
1859-1893
1861-1876
1863-1879
1865-1875
1867-1875
1868-1877
1867-1890
1867-1874
1868-1891
1870-1890
1870-1877
1871-1882
1872-1897
1872-1874
1872-1877
1873-1874
1873-1883
1874-1907
1874-1891
1874-1877
1881-1903
1883-1945
1886-1888
1891-1922
1896-1922

Production figures for two furnaces

Source: LaFayette 1977:49.
MUN - Munising
MAR - Marquette

N/A - Other than Lake Superior port

 

41,997

30,059

637,299

55,608
57,573
40,202
41,531
23,312

31,048

229,228

41,857

11,346
50,706

8,650
59,553
83,500
93,579

8,209
58,349
73,829

3,700

Port of

Production Shipment

N/A
MAR
MAR
N/A.
Christmas
MAR
MAR
MAR
Christmas
N/A
N/A-
MAR
MAR
MAR
N/A
N/A
N/A
N/A
N/A

222

Table 3. Iron ore shipments by range

Year Marquette Menominee Vermillion Gogebic Mesaba
1854 3,000

1855 1,449

1856 36,343

1857 35,646

1858 65,878

1859 68,832

1860 124,401

1861 49,909

1862 124,169

1863 203,055

1864 243,127

1865 236,208

1866 278,796

1867 473,567

1868 491,449

1869 617,444

1870 830,940

1871 779,607

1872 900,901

1873 1,162,458

1874 919,557

1875 891,257

1876 992,764

1877 1,010,494 4,593

1878 1,033,082 78,028

1879 1,130,019 245,672

1880 1,348,010 524,735

1881 1,597,834 727,171

1882 1,829,394 1,136,018

1883 1,305,425 1,047,415

1884 1,548,034 895,634 62,124 1,022

1885 1,480,422 690,435 225,484 119,860

1886 1,627,380 880,006 304,396 753,362

1887 1,851,414 1,193,343 394,252 1,322,878

1888 1,923,737 1,191,101 511,953 1,437,096

1889 2,642,813 1,796,754 884,682 2,008,394

1890 2,993,664 2,282,237 880,014 2,847,810

1891 2,512,242 1,824,619 894,618 1,839,574

1892 2,666,856 2,261,499 1,167,650 2,971,991 4,425
1893 1,835,893 1,466,197 820,621 1,329,385 613,620
1894 2,049,107 1,137,949 948,513 1.809.468 . 1,793,052
1895 2,097,838 1,923,798 1,077,838 2,547,976 2,781,587
1896 , 2,604,221 1,560,467 1,088,090 1,799,971 2,882,079
1897 2,715,035 1,937,012 1,278,481 2,258,236 4,280,873
1898 3,119,000 2,522,000 1,265,000 2,498,000 4,614,000
1899 3,738,000 3,301,000 1,772,000 2,799,000 6,614,000
1900 3,479,000 3,261,000 1,656,000 2,877,000 7,810,000
1901 3,247,000 3,619,000 1,786,000 2,938,000 9,005,000

Iron ore shipments by range continued

Table 3.

Year Marquette
1902 3,865,000
1903 3,040,000
1904 2,852,000
1905 4,236,000
1906 4,057,000
1907 4,388,000
1908 2,414,000
1909 4,253,000
1910 4,393,000
1911 2,836,000
1912 4,203,000
1913 3,968,000
1914 2,494,000
1915 4,106,000
1916 5,410,000
1917 4,874,000
1918 4,354,000
1919 2,992,000
1920 4,608,000
1921 1,117,000
1922 2,818,000
1923 3,892,000
1924 3,175,000
1925 4,198,000
1926 4,435,000
1927 4,148,000
1928 4,299,000
1929 5,410,000
1930 3,634,000

Sources:

Menominee

4,613,000
3,750,000
3,075,000
4,495,000
5,110,000
4,965,000
2,679,000
4,875,000
4,238,000
3,911,000
4,711,000
4,967,000
3,222,000
4,983,000
6,365,000
6,046,000
6,379,000
4,447,000
6,569,000
1,584,000
4,079,000
4,855,000
3,837,000
5,270,000
5,946,000
5,213,000
4,842,000
5,645,000
3,609,000

223

Vermillion

2,084,000
1,677,000
1,283,000
1,677,000
1,793,000
1,685,000

842,000
1,109,000
1,203,000
1,089,000
1,845,000
1,567,000
1,017,000
1,734,000
1,947,000
1,531,000
1,193,000

929,000
1,007,000

869,000
1,212,000
1,279,000

978,000
1,438,000
1,586,000
1,548,000
1,671,000
1,874,000
1,885,000

Mansfield 1899:566; Hickok 1938:308

Gogebic

3,659,000
2,939,000
2,399,000
3,706,000
3,642,000
3,633,000
2,700,000
4,088,000
4,316,000
2,603,000
5,006,000
4,532,000
3,569,000
5,478,000
8,490,000
7,980,000
7,937,000
6,230,000
8,763,000
2,337,000
6,221,000
6,580,000
5,160,000
7,068,000
7,537,000
6,386,000
6,540,000
7,624,000
5,064,000

Mesaba

13,331,000
12,894,000
12,157,000
20,159,000
23,821,000
27,492,000
17,258,000
28,168,000
29,200,000
22,099,000
32,045,000
34,040,000
21,468,000
29,757,000
42,526,000
41,441,000
40,399,000
32,004,000
37,150,000
16,360,000
28,064,000
41,806,000
29,142,000
35,890,000
38,251,000
32,976,000
35,399,000
43,008,000
31,067,000

Table 4.

224

Grain and flour shipments-Lake Superior 1870-1911

Year

1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911

Grain Shipment

10.0
50.0
29.5
72.3
37.8
43.4
70.6
50.1
63.6
104.7
134.5
114.0
125.1
195.8
374.1
470.1
589.9
714.6
614.5
546.7
543.7
1193.4
1276.5
1371.8
1089.3
1619.8
2666.2
2265.4
2600.4
2591.1
899.9
1090.9
1193.2
958.0
986.0
1174.1
1580.5
1626.2
1570.3
2067.0
1161.7
1116.0

Shipments given in thousand tons.

Flour Shipment

kkmomNu
o o o o
O-QOO‘DN

uw—ou—o—o

39.1

37.8

49.6

57.6

66.5

37.8

75.6
137.3
158.4
193.5
172.9
240.9
245.1
356.2
415.8
596.0
816.3
986.2
979.2
977.1
981.3
855.6
782.5
688.0
633.3
743.7
564.9
442.7
454.1
583.8
515.1
402.8
455.4
455.9
459.1

Table 5.

Sources:

225

Coal Shipment--Lake Superior 1855-1919

Year

1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887

Mansfield 1899:194; MacElwee 1921:242; Williamson 1977:234-235

Amount

1,414
3,968
5,278
4,118
8,884

11,507
11,346

7,805
11,282

19,915
22,927
25,814
27,850
15,952
46,798
80,815
96,780
61,123
101,260
124,734
91,575
91,856
110,704
170,501
295,647
430,184
714,444
706,379
894,991
1,009,999

1,352,987

Year

1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919

Amount

2,105,041
1,629,197
2,176,925
2,507,532
2,904,266
3,008,120
2,797,184
2,574,362
3,023,340
3,029,172
3,776,450
3,940,900
2,896,100
4,240,000
4,300,800
6,022,400
5,935,400
5,643,200
7,637,500
9,683,400
8,227,900
8,121,200

13,513,727

15,332,876

14,931,594

18,662,938

14,487,221

13,357,058

16,123,119

18,298,853

17,981,610

13,874,951

APPENDIX B

VESSEL DATA

226

 

mm ouuosoumz eOHmHHHou cme .NN mHah swamp amoum poo: oeuoomc z a
on nauseous: weresosm mmmH .a mHsh swamp amour onoumn and:
x mm muuosoumz weHoenoum meH .ON .>oz omuoe\soeooeom omeoeoem
on weHmHesz eOHmHHHoo meH .oN .wse uoHHooouo uoaoo
x on ouuosouoz onoesoum mNmH .nN .uoom poHHooouo eoHeD
x on nauseous: weHoesosm NNQH .ms< nonconom oonuoamo
x mm masseuse: weHoesouw NNwH .NN .>oz owumn\uoeooeom ensumm
x mm ouuosouoz weHoenouw NNmH .NN .>oz owume\uoeooeom souHosw
N on venomous: wsHoeoouw aomH .e .089 noeooeom ouomon
x mm ouuosouoz meHoesoum oowH .m .>oz noeooeom oHoeh< 3 3
sensuous: onuooesom momH .uoom uoeooeom msoHosz
on ouuosouoz weHoenouw wowH he: Hoeooeom oneo oeHoeo
on sensuous: eonHHHoo NomH .m .w:< uoeoonom oHoHno
mm masseuse: meHuooesom mmmH mesh uoHHooouo meoHoeH
AooON
eommHomv
neomoum umoH eHmHuo poem «0 mmoH
manoum eOHumUOH mo uuom GOHuHoeou we soon ooze Hommo> mamz Howmo>

mummOH H<HOH I mmo zomH

227

 

mm esoexes eOHmHHHoo HooH .oN mesh owumn xomeones ouoaowmm
mm muoonom 039 weHuooeeom oomH .N as: owuoo\uoeooeom emumHHom m
x o esoexee weHoenouw mowH .MH .oon omens somooHoes
x mm masseuse: weHoesouw NmmH .Om .>oz omuon\uoeooeom owHom seamen
on euoHsa weHmesouw «owH .NN .uoom seasons woos moemeon peace
on ouuosoumz eonHHHoo NomH .oH .uoom anemone oooB oeeoH>
x HH H3 .oemHem4 meHuooesom oomH .NH .uoo owume\uoeoosom opouaoo
x mm H: .ssssea< meansessom amen .m .soo mmsss\smaooeum Haasssssum
x mm H3 .oeoHeo< weHuopesom ommH .m .uoo amuse amour woos uoaommom
eBoexes GOHmHHHoo mwwH .MH mesh amuse amour woos ouooz euHem
x mm H3 .oeoHem< weHuooeoom waH .N .uoom owumo\uoeooeoo snomez
x Na H3 .oemHem< meHuooesow ome .NH .>oz owuon\uoeooeom oeuoosH
x mm chansons: onuooesom ome .ON .uoo owuooxuoeooeum eschew
Nm choose: ose weHuooosom mmmH .mN .uoom owuoo\uoeooeoo uoHooe3 3 m
Aoeou
noomHomv
unomoum umoH eHwHuo moon mo mmoq .
mahoom eoHuoooH mo uuom eOHqueoo we some some Homoo> oasz Hommo>

AneezoezouV mmmmoa gases 1 mac zomH

228

 

eusHsn eaoexes momH .N .uoom Hoamoum oooB oumoH
x eusHsa meHuooesom momH .N .uoom omume\soeooeom euuoemom o>HHo
x Nm eusHsa meHuomesow momH .N .ueom omuoo\uoeooeom oHuououm
x No uoHnoosm meHoesoum mooH .N .uoom seasons Hooum meo>om
on ouuooouoz meHoesoum comH .OH .uoom seasons woos mqum
x mm H3 .ocoHem< meHuooesom momH .mN .uoo seasons poo: season .3 EoHHHH3
N Na H3 .meoHem< meHuoossom momH .m .uoo seasons pooa ouuosoumz
N on sOHuoosm meHuooeeom mooH .oH .uoom someone pooz noxumm 4 <
N ma H3 .oeoHem4 meHuooe30m momH .mH .uoom omume\soeooeoo uanHeooz
we eusHsa meHuooesow NomH .NH .>oz seasons woos ouoHHm3 uuoeom
x Ho saoexes eonHHHoo NooH .uoo amuse someones
«a eusHan eonHHHoo NomH .N mesh seasons sooooHoes eomHH3 seesaw
x mm eaoexes eoHoHHHoo HomH .N .uoo seasons poo: among 3 z
x mm esoexe: meHuooesow HomH .N .uoo omumo\uoeooeom eomHnon
Aoeom
eoooHomv
neoooum umOH sHmHuo mmoa mo mmoq
mahoum eOHuouOH mo uuom eoHuHoeoo we use: some Hommo> mamz Hommm>

anmszsezouV summon n<eoa 1 use

ZOMH

229

 

on muonsmm 039 eonHHHoo ONmH .ON .m:¢ seasons Hooum auHo uoHsoonm
x on H3..oeoHem¢ meHuopesom onH .m an: uoaooum moos\eonH mouHM m m
N am mandamus: meHuooenom NHoH .m .>oz uoaooum Hooum euHEm .m mason
nauseous: meHuove50m NHoH .NN .uoom anemone woos ommHHHso
mm neoexee eonHHHoo mooH .NH hHsn seasons House oHsou .m snow
mm HOHHoonm meresoum mooH .NN .uoom Hoamoum woos ouoemoz
mm uoHuooom meHsooenom NooH .HH .uoo uoamoum Hooum msuoho
on amHHHH3 usom ouHm NomH .oN mesh monsoon poo: meeosmeoumm
AoeoN
sommHomv
aeomoum unoH eHmHuo when we mmog
mahoum soHuoUOH mo whom eOHuHoeou we some some Hommo> oaoz Hommo>

AnsszHeonV

mummoa H<HOH I mmo zomH

230

 

om eBoexes meHoesoum mooH .m .m:< noamoum xoonones HopeHweumm
on uoHuoesm meHoesosm momH .QN .uoo seasons Hooum seesaw: .m HoHeoa
on neoexes meHoesoum oomH .nN he: seasons House mason:
x mo esoexes msHoeeoum nooH .mN .>oz uoaooum Hooum zouoo .m emHHHH3
mH nueHnn meHoesoum momH .N .>oz omumn Hooum meHoz
x mH euoHsa meHoesosm nomH .N .>oz seasons someoHoaa uoHHomoxoom semen
as H: .esmssma «use moms .nm .wasm swsms\umsooeom goose .m assess
no sameness: GOHmHHHoo ommH .oN .so4 omumexuoeooeom oHuoemmz
on eBaexes meHeeeoum mmmH .oH seam Hoeooeom mundane:
x No ouuosouoz meHeesosm ommH .cN .>oz uoeooeom UHHosoom
an enemas: meHoesoum ommH .Om .mn< omuoexuoeooeom esoum ouuomom
mm ozoexes meHoesoum ommH .om .msm soHHooouo onm u m
x mm ouuonoumz meHoeooum onH .uoo Hoeooeom mosauom
an ouuosopmz osHm momH .NH mesh uoHHooouo momma can no noose
Assam
nommHomv
anemone umoH eHmHuo mmOH mo mmoH
mEMOum eOHumooq mo uuom eoHuHoeoo me some ooha Hommo> oeoz Hammo>

QMU<>H<m I mmo ZOMH

231

 

«a HOHuoosm eOHoHHHoo quH .NH he: seasons Hooum some: one uo>oom
an nauseous: msHoesoum HNmH .ON .>oz seasons Hooum oHomoem m m
on H3 .oemHem4 meHvesoum «HmH .m as: seasons Hoouo women .m omuooo
N am eusHsn meHoesoum MHoH .OH .>oz Hoamoum Hooum HHoeuumm .6 seem
x so eusHso meHoesoum mHoH .OH .>oz seasons Homum eoeeHeouom H m
x on mucous: one meHoesosm MHmH .m .ms4 seasons Hooum ooHo3 o H
on noHuoosm meHoesoum CHoH .mH .uoo seasons Hooum osmHesoz .u aoHHHH3
on sensuous: meHoeooum oHoH .oN hHsm umaooum Hooum huHo euHeoN
x cN susHsn meHoeooum oomH .mH .>oz seasons Hooum ouommemum
x mH ascends meHoeoonm momH .HH .uoo seasons someones uo>oua .o anon
on uo>Hm onmm meHoesoum momH .m .ms< omumo Hooum ouofimmmm
Assam
nommHomv
someonm umoH eHmHuo smog mo smog
mauoum eoHumooq mo uuom eoHuHoeoo we more some Hommo> oaoz Hommo>

amazes—48V $353

I mmo zomH

232

 

x useuu< uuom eaoexes NomH .HN .>oz soamoum assessmeemm

a on assess secesss Hess .sn .saom sssemasse amass sesame

x oH eusHsa meHoesoum mmmH .NH .uoo soamoum poo3 eHoemHno mason

x on eusHsa meHoesoum mamH .mH .uoom seasons woos owmsoHoo

mm assess «use some .AN .uaa smammsm e663 sa>smm

on eusHsn msHuooesom mmmH .N .>oz swamoum woos mHsommHz

x mm eusHsa meHsmoenom nomH .HN .uoom omume\uoeooeom xooumaoo 3 H

mm eusHsa eonHHHoo HamH .eH .>oz seasons woos senses Hossmm

x mm euann meHvesoum emmH .HN .>oz seasons woos hoHuoz

eusHsn meHuopenom mmmH .OH .>oz uoHHooouo woos moumHemz

He eusHoa meHuooenom MNmH .MN .uoo seasons poo: o>meou

x mm assess seassessoe News .NN .>oz ussoosum enosmasu .0 assumes

x no eueHsn meHoeooum NNmH .NN .>oz noeooeom ozoum o 3

x mm eusHsa meHoesoum HNmH .NH .uoo nonooeom uo>on
Aoeom
nommHomv

unomoum umoH eHmHuo mmoH mo mmoq
mahoum eOHumUOH mo uuom eOHuHoeoo mo sumo some Hommo> mamz Hmmmo>

mummOH H<HOH I ZH<MU

 

‘5w

233

 

N 0N EMHHHH3 uuom meHmesoum «NmH .H .>oz HousmHoum omoxomo somHeoHo

N on euoHsa meHoesoum oHoH .oN .>oz seasons omoHH

N eusHsn esoexes mHmH .m .>oz soamoum ouHmoeaoo mono seem

N OH amHHHH3 .um meHvesoum mHmH .o .>oz seasons Hooum eonmsoo .< noumoeu

a Na assess «use mass .mn .>oz seasons e663 names: .A esstaa

no eusHea meHsooesom mHmH .oH .uoom seasons eouH . oonso

N NH useuu< uuom possessom momH .nH .>oz seasons Hooum osmuuo

N HH useuu< uuom meHeesoum oomH .o .009 seasons poo: ecumeoz

N mN aoHHHH3 uuom meHoeooum oomH .mN .>oz seasons woos oHomHow

N 0N emHHHH3 uuom merenoum cooH .qH .>oz seasons woos opossumuum

x on assess sustenance moss .mm .>oz seasons case .= ssH

N No eusHsa meresoum momH .oN .uoo seasons noes seaweeds:
Aoeom
dommHomv

newsman umoH :HmHuo mmoH mo mmoH
mauoum eOHumooA me upon eoHqueoo mo mama mama Hommo> oemz Hommo>

Ansaznszoov mmmmoa pesos I

zH<MU

234

 

x on £38 assesses» $3 .3 as: $88: H636 283 firms
N on :3oexe: meHoesoum NHmH .oH .>oz Monsoon Hooum amemeHuuoz amHHHHB
om euann osHm HomH .om mesh seasons woos eoe<
N on eusHsn meHoeooum mmmH .NN .>oz souemHouw .mxo Hooum who useus<
N ON nusHsa meHoesoum nomH .mN .uoo seasons Homum eoHusueoo
N mm eusHso meHuooesom nmmH .cN .uoom seasons woos oeoueoz
x a... £38 assesses» an: .3 $62 69:3 831: 323
N on euana meresoum ommH .mH .>oz uoamoum poo: oooHHo3 unoeom
N useuu< uuom msHoesoum ammH .oN seam uoHHoooso emHmuoou
Ameom
nomoHomv
ueomonm umoH eHmHuo mmoq mo mmoH
mauoum eOHumooq mo uuom eoHuHoeoo me some ooNH Hommo> oemz Hommo>

nmw<>q<m I zH<MU

235

 

N on ozoexes meHoesoum momH .NN .uoo omuon\soeooeom sea .< oHsNHH
N mm ekoexes meHoesoum NcmH .HN .uoom soaoouo woos onaHz on<
N as eBoexes meHoesoum oomH .HN .>oz Hoaoouo oooB oaouuom
NH esoexes meHuooesom cooH .mN .uoo omsoe\uoeooeom eHmHm
N mm eaoexes meHoesoum oomH .m .uoo uoeooeoo euooomom
N Nn esoexes moHuooenom mmmH .mH No2 omuoo\soeooeuo eomHoz
mo eaoexe: meHnooesom NmmH .N .uoo omuoe\uoeooeoo oooHoue<
N mm eaoexe: meHoesoum mmmH .QN .uoo uoeooeom omuoou
N on eaoexe: meHuooenom HomH .oN .uoom uoeooeom Boson xeoum
N mm peoouoo3 meHuopesom HmmH .m No: soeooeoo oueoHu<
N on esoexeo meHvesoum NmmH .MN .uoo uoeoosom eoeuoem omuoou
N on saoexen meHoesowm ommH .NH .>oz Honooeum ooHHOHm
N on esoexe: msHoesonm mmmH .m .uoom uoeooeoo HHom enema
on ozoexos meHoeoosm mNmH .>oz uoeooeoo moose .H seem
Aoeom
eoomHomv
ueomonm umOH eHmHuo mooa mo moon
mauoum eoHuooOH mo uuom :OHuHoeoo mo ouon ooze Hoomo> oaoz Hoomo>

mmmmOH H<HOH I H<OU

236

 

N mm e3oexe= meHuooesom NNmH .om .>oz uoaooum Hooum nouneoHomz
N Nm eaoexes meHoeooum aHmH .aH .>oz uoxoo: sonasH oHoeeoN m 3
NH ozoexes pouooeoow mHoH .oN No2 Monsoon poo: someHmoHeom oeoeHsom
N on obsess: meHoesoum mHmH .H .uoo Hoaooum poo: ooHooum oHoo
mH eBoexes GOHmHHHoo NHmH .N .ms¢ uoaoouo Hooum NoHNoo moaoh
N m eBoexes meHoenoum HHmH .mN NHsh Hoaoouo poo: Nooeeoooom
ssoexes eOHmHHHoo HHmH .OH NHsm Hoaooum Hooum HHoeoqu seem
:Boexes ouHm OHmH .mN .uoo soaoouo woos oaoemeoq
Nm oeHouoa eaoexes mooH .H .ooa soaoouo Hoouo eooaoHo z a
AoGON
eoomHomv
ueoooum umoH eHmHuo mmoH mo smog
mahoum eOHuoooH mo uuom GOHuHoeou mo ouon ooze Hommo> oaoz Hommo>

Anssznezoov mammos o<eoe 1

237

 

N Nn czoexes meHoesoum mHmH .om .uoo noaoouo Hooum cooHo>

mN eaoexeo meHoeooum. mooH .mN mHsh soaoouo Hooum aoeoooz .m oH>on

N mm sneeze: meHoesoum mooH .aH .uoo omuon\uoeooeom eHeuoz .m eooHom

«a esoexes eoHoHHHoo mooH .qH .uoo uoaooum poo: oueoaosoom

N Nm esoexes meHoesoum NomH .mN .uoo soaooum Hooum oeoxoom

as esoexes eonHHHoo NomH .N oesm uoaoouo woos NoHoom .0 omnooo

N om muoexe: meresoum mmmH .NN .>oz uoaoouo woos moans

N ma assess: meHoesoum momH .mH No2 Hoemouo Hooum omo>

N Nm esoexes meHoesoum NomH .oN .>oz soxoo: HooanH woos mHHoeeom m m

«a :Boexes meHoeooum nmmH .OH .uoo Hoeooeoo Hosonm ooHsu

N on :3oexes meHuoocsom NNmH .mH .uoom uoeooeom omoz N <

N on :Soexeo meHnooesom NNmH .mH .uoom nonooeom mononom o m
AoGON
eommHomv

ueomoum uooH eHmHuo mooq we smog
mauoum eoHuoooH mo uuom eoHuHoeou mo ouon ooze Hommo> oaoz Hoomo>

nmu¢>q<m I H400

APPENDIX C

APPENDIX C

This appendix comprises the series of tables from chapter 6
which provide raw data on frequency distributions, etc. In order to
link these tables with their respective hypotheses they have been
numbered in accordance with the hypotheses they concern, rather than
in simple numerical sequence. Tables from hypothesis 1 will be listed
1a through 1d, hypothesis 19 as 19a through 19h, and so forth. Some
hypotheses will not have tables in this appendix so numerical gaps will

be present at some points.

238

239

Table 1a. All commodities - salvage and total loss combined

 

 

Poisson Zone Frequency Poisson Zone Frequency

5 1 41 2
6 1 43 1
7 1 44 5
10 1 46 1
11 2 47 4
12 ' 2 48 4
13 1 49 1
16 1 52 2
17 1 53 1
19 4 54 12
20 3 55 3
26 1 56 8
28 1 57 6
29 1 58 15
30 2 59 15
32 2 6O 1
35 5 62 1
36 15 64 2
37 1 65 1
38 2

Table 1b. Iron - salvage and total loss combined

 

 

Poisson Zone Frequency Poisson Zone Frequency
6 1 47 4
11 1 48 2
19 3 54 S
20 1 55 2
32 1 56 7
35 2 57 2
36 9 58 11
38 1 59 10
41 1 61 1
44 2 64 2
46 1 65 1

240

Table 1e. Grain - salvage and total loss combined

 

 

Poisson Zone Frequency Poisson Zone Frequency
7 1 36 4
10 1 38 1
11 1 41 1
12 1 43 l
16 1 45 1
19 1 54 2
20 2 57 1
26 1 58 1
28 1 59 4
30 1 60 1
32 1 62 1
35 1

Table 1d. Coal - salvage and total loss combined

 

 

Poisson Zone Frequency Poisson Zone Frequency
5 1 48 2
12 1 49 1
13 1 52 2
17 1 53 l
29 1 54 5
30 1 55 1
35 2 56 1
36 2 57 3
37 1 58 2
44 3 59 1

Table 2a. 1855-1879 - Iron

 

 

Poisson Zone Frequency Poisson Zone Frequency
54 3 58 4
55 2 59 1
56 3

241

Table 2b. 1855-1879 - Grain

 

 

Poisson Zone Frequency Poisson Zone . Frequency
41 1 58 1
43 1. S9 1

Table 2c. 1855-1879 - Coal

 

Poisson Zone Frequency
36 1
54 2

Table 2d. 1880-1904 - Iron

 

 

Poisson Zone Frequency Poisson Zone Frequency

6 1 47 2

11 1 48 1

35 2 56 3
36 1 57 1
38 1 58 5
41 1 59 6
44 1 62 1
46 1 65 1

Table 2e. 1880-1904 - Grain

 

 

Poisson Zone Frequency Poisson Zone Frequency

7 1 54 2

19 1 57 1

20 1 58 1
30 1 59 2
35 1 60 1
36 2 62 1

1

38

242

Table 2f. 1880-1904 - Coal

 

 

Poisson Zone Frequency Poisson Zone Frequency
30 1 55 1
44 2 57 1
48 2 58 1
52 1 59 1
54 3

Table 2g. 1905-1929 - Iron

 

 

Poisson Zone Frequency Poisson Zone Frequency

19 3 54 2
20 1 56 1
32 1 57 l
36 9 58 1
44 1 59 .3
47 2 64 2
48 1

Table 2h. 1905-1929 - Grain

 

 

Poisson Zone Frequency Poisson Zone Frequency
10 1 . 28 1
11 1 32 1
12 1 36 2
16 1 45 1
20 1 59 l
26 1

Table 2i. 1905-1920 - Coal

 

 

Poisson Zone Frequency Poisson Zone Frequency
5 1 44 1
12 1 49 1
13 1 52 1
17 1 53 1
29 1 56 1
35 2 57 2
36 1 58 1
1

37

243

Table 6a. Marquette

 

 

Poisson Zone Frequency Poisson Zone Frequency
54 5 59 2
55 2 62
56 6 65 1
58 6

Table 6b. Ashland

 

 

Poisson Zone Frequency Poisson Zone Frequency
11 1 47 3
35 2 48 1
36 2 58 2

Table 6c. Two Harbors

 

Poisson Zone Frequency
36 1
38 l
57 1
59 1

Table 6d. Duluth/Superior

 

 

Poisson Zone Frequency Poisson Zone Frequency
16 1 45 1
19 3 46 1
20 1 47 1
30 1 54 2
32 2 56 1
35 1 57 2
36 7 58 3
38 1 59 3
41 1 60 1
43 1 62 1
44 2 64 2

244

Table 6e. Port Arthur

 

 

Poisson Zone Frequency Poisson Zone Frequency
7 1 20 1
10 1 26 1
11 1 28 1
12 1 59 1

Table 10a. Grounding

 

 

Poisson Zone Frequency Poisson Zone Frequency
5 l 44 1
6 1 47 1
7 1 48 2

10 1 49 1
11 1 52 2
19 4 53 1
20 3 54 8
26 1 55 3
28 1 56 6
29 1 57 2
30 2 58 7
35 1 59 4
36 13 62 2
37 1 64 1
43 1

Table 10b. Foundering

 

 

Poisson Zone Frequency Poisson Zone Frequency
11 1 45 1
12 2 46 1
l6 1 47 2
17 1 48 2
32 1 54 3
35 4 56 1
36 1 57 3
38 2 58 6
41 1 59 3

245

Table 10c. Collision

 

Poisson Zone Frequency
13 1
41 1
44 4
56 1

Table 10d. Fire

 

Poisson Zone Frequency
32 1
47 1
54 1

Table 13a. Schooner

 

Poisson Zone Frequency
35 1
43 1
44 1
54 7
55 2

Table 13b. Schooner-barge

 

Poisson Zone Frequency
11 1
17 1
32 1
35 1
36 2
38 2
47 2
48 2

Poisson Zone
58

59
65

Poisson Zone

 

58
59
60

Poisson Zone

56
58
59
62

Poisson Zone

 

53
54
55
57
58
59
65

Frequency
1

1

Frequency

Frequency

Frequency.

1
1
l
2
8
1
l

246

Table 13c. Wooden steamer

 

 

Poisson Zone Frequency Poisson Zone Frequency

5 1 44 2
1 1 1 46 1
12 1 47 1
19 1 49 1
26 1 54 1
28 1 56 4
30 1 57 1
32 1 58 5
35 2 59 5
36 3 6O 1
41 1 62 1

Table 13d. Steel steamer

 

 

Poisson Zone Frequency Poisson Zone Frequency
10 1 44 1
12 1 47 1
13 1 48 2
16 1 52 1
20 2 54 2
29 1 56 1
35 1 57 2
36 6 59 3
37 1 64 2

Table 13e. Miscellaneous vessels

 

 

Poisson Zone Frequency Poisson Zone Frequency
6 1 45 1
7 1 52 1
19 3 54 1
20 1 56 1
30 1 57 1
33 1 58 1
36 4 59 3
1

44

Vessel capacity

247

 

 

 

Table 15.
Year Tons
1855 500
1856 550
1857 600
1858 650
1859 700
1860 750
1861 800
1862 850
1863 900
1864 950
1865 1000
1866 1050
1867 1110
1868 1150
1869 1200
1870 1200
1871 1350
1872 1500
1873 1650
1874 1800
1875 1950
1876 2100
1877 2250
1878 2400
1879 2550
1880 2700
Table 19a.

Total loss

Poisson Zone

 

6
11
32
35
36
38
41
44
46

Year Tons
1881 2400
1882 3000
1883 3215
1884 3430
1885 3645
1886 3860
1887 4075
1888 4290
1889 4505
1890 4720
1891 4935
1892 5150
1893 5365
1894 5580
1895 5800
1896 6280
1897 6760
1898 7240
1899 7720
1900 8200
1901 8740
1902 9280
1903 9820
1904 10360
1905 10900
Frequency

 

47
48
54
55
56
57
58
59

Year Tons
1906 11440
1907 12000
1908 12065
1909 12130
1910 12195
1911 12260
1912 12325
1913 12390
1914 12455
1915 12520
1916 12585
1917 12650
1918 12715
1919 12780
1920 12845
1921 12910
1922 12975
1923 13040
1924 13105
1925 13170
1926 13235
1927 13300
1928 13365
1929 13430
1930 13495
Poisson Zone

Frequency

\l—NO‘—U—'U

248

Table 19b. Iron - salvage

 

 

Poisson Zone Frequency Poisson Zone Frequency

19 3 55 1
20 1 56 1
36 7 59 3
44 1 62 1
47 1 64 2
48 1 65 1
54 2

Table 19c. Grain - total loss

 

 

Poisson Zone Frequency Poisson Zone Frequency
10 1 36 3
11 1 38 1
12 1 41 1
16 1 43 1
19 1 45 1
20 1 57 1
26 1 58 2
28 1 59 3
32 1 62 1

Table 19d. Grain - salvage

 

 

Poisson Zone Frequency Poisson Zone Frequency
7 1 36 l
20 1 54 2
30 1 59 1
35 1 60 1

Table 19e. Coal - total loss

 

 

Poisson Zone Frequency Poisson Zone Frequency
5 1 49 1
12 1 54 3
13 1 55 1
17 l 56 1
35 2 57 3
36 2 58 2
48 1 59 1

249

/A

Tab 2 19f. Coal - salvage

 

 

Poisson Zone Frequency Poisson Zone Frequency
29 1 48 1
30 1 52 2
37 1 S3 1
44 3 54 2

Table 19g. All commodities - total loss

 

 

Poisson Zone Frequency Poisson Zone Frequency
5 1 41 2
6 1 43 1

10 1 44 1
11 2 45 1
12 2 46 1
13 1 47 3
16 1 48 2
17 1 49 1
19 1 54 6
20 1 55 2
26 l 56 7
28 1 57 6
32 2 58 15
35 4 59 11
36 7 62 1
38 2

Table 19h. All commodities - salvage

 

 

Poisson Zone Frequency Poisson Zone Frequency

7 1 52 2
19 3 53 1
20 2 54 6
29 l 55 1
30 2 56 1
35 1 59 4
36 8 60 1
37 1 62 1
44 4 64 2
47 1 65 1
48 2

250

Conclusion. Iron - storms

 

 

Poisson Zone Frequency Poisson Zone Frequency
11 1 48 1
19 2 54 3
20 1 55 2
32 1 56 2
35 2 57 1
36 2 58 9
41 1 62 1
47 3 64 2

Conclusion. Grain - storms

 

 

Poisson Zone Frequency Poisson Zone Frequency
10 1 32 1
ll 1 36 4
12 1 38 1
16 1 43 1
19 1 54 2
20 2 57 1
26 1 58 1
28 1 59 2
30 1 62 1

Conclusion. Coal - storms

 

 

Poisson Zone Frequency Poisson Zone Frequency

5 1 53 l

30 1 54 5
35 2 55 1
36 1 56 1

37 1 57 2
48 1 58 2
49 1 59 2

2

52

251

Conclusion. .All commodities - storms

 

 

Poisson Zone Frequency Poisson Zone Frequency
5 1 41 1
6 1 43 1
10 1 47 3
11 1 48 3
12 1 49 1
16 1 52 2
19 3 53 1
20 3 54 10
26 1 55 3
28 1 56 3
30 2 57 4
32 2 58 11
35 5 59 3
36 7 62 2
37 1 64 2
38 1

REFERENCES CITED

REFERENCES CITED

Bald, Frederick E. 1961. Michigan iu_Four Centuries. Harper, New
York.

 

Barry, James P. 1973. Ships of the Great Lakes: 300 Years of
Navigation. Howell-North Books, Berkeley.

 

Bass, George F. 1972. ‘A History of_Seafaring bas d ou_Underwater
Archaeology. walker & Co., New York.

 

Bayer, E. P. 1938. Grading to Cargo Sepcifications. In Lake Superior
Iron Ores, Chris Hickok ed. Lake Superior Iron Ore Assoc.,
Cleveland.

Berry, Brian J. and Duane F. Marble eds. 1968. Spatial Analysis:
.A leader Eu Statistical Geography. Prentice-Hall, Englewood
Cliffs, New Jersey.

 

Clarke, David L. 1968. Analytical Archaeology, Methuen, London.

Dacey, Michael F. 1968. Modified Poisson Probability Law for Point
Pattern More Regular Than Random. In Spatial Analysis: A
Leader iu_Statistical Geography, Brian Berry and Duane Marble
eds. Prentice-Hall, Englewood Cliffs, New Jersey.

 

Doran, J. E. and F. R. Hodson. 1975. Mathematics and Computersiyu
Archaeology. Harvard University Press, Cambridge, Mass.

Erickson, Bernard E. 1969. The Evolution of Great Lakes Ships. Part
II, Inland Seas 25(3):l99-212.

Evans, Henry 0. 1942. Iron Pioneer: Henry W, Oliver 1840-1904. E.
P. Dutton & Co., New York.

 

Garrety, John A. 1971. The American Nation - A_History g£_the United
States 53 1877. Harper & Row, New York.

 

Hatcher, Harlan. 1950. ‘A Century 1 Iron and Men. Babba-Merrill
Co., New York.

 

Heden, Karl. 1966. Directory of Shipwrecks of_the Great Lakes.
Bruce Hamphries, Boston.

 

 

252

253

Hickok, Chris N. (ed.) 1938. Lake Superior Iron Ores. The Lake
Superior Iron Ore Association, Cleveland.

 

Hodder, Ian and Cline Orton. 1976. Spatial Analysis iu Archaeology.
Cambridge University Press, Cambridge.

Holland, Francis R. 1972. American Lighthouses - Their Illustrated
'HistorySince 1716. Stephen Green Press, Brattleboro, Vt.

 

Hulse, Charles A. et a1. 1977. Archaeological Survey of Rock Kilns
and Finn Hollow (20 AR 10), Alger Co., Michigan. Archaeological
Survey Report No. 24, MSU Museum Publication.

Jarchow, Merrill. 1948. King Wheat. Minnesota History, 29(1):l-28.

Johnson, Gilbert R. 1948. United States-Canadian Treaties Affecting
Great Lakes Commerce and Navigation. Inland Seas 4(2):ll3-ll9.

LaFayette, Kenneth D. l977.F1amiug_Brands: Fifty Years of Iron
Making_ in the Upper Peninsula of Michigan 1848-1898. —Northern
Michigan University Press, Marquette.

Life Saving Service, United States 1877-1914. Annual Reports, District
No. 9.

Mac Elwee, Roy S. 1921. Economic Aspects of_the Great Lakes - SE.
Lawrence Ship Canal. Ronald Press, New York.

 

 

MacGibbon, Duncan. 1932. The Canadian Grain Trade, Macmillan Co.,
Toronto.

 

Mansfield, J. B. (ed) 1899. History of the Great Lakes. 2 Vols.,
J. H. Beers & Co., Chicago.

Mueller, James W. Ed. 1975. Sampling iu Archaeology. University of
Arizona Press, Tucson.

Murray, C. B. 1938. Classification and Sampline of Ore. In Lake
Superior Iron Ores, Chas. Hickok ed. Lake Superior Iron Ore
Assoc. , Cleveland.

 

Neu, Irene D. 1953. The Building of the Sault Canal 1852-1855. The
Mississippi Vallengistorical Review, 40(l):25-46.

Nute, Grace Lee. 1944. Lake Superior. Bobbs-Merrill Co., New York.

 

O’Brien, T. Michael. 1976. Guardions gf_the 8th Sea - A_history of
the United States Coast Guard ou the Great Lakes. U. S. Coast
Guard, Washington, D. C.

 

 

Odle, Thomas D. 1951. The American Grain Trade of the Great Lakes,
1825-1873. Part I. Inland Seas 7(4):237-245.

 

254

Odle, Thomas D. l952a. The American Grain Trade of the Great Lakes,
1825-1873. Part II. Inland Seas 8(1):23-28.

 

Odle, Thomas D. l952b. The American Grain Trade of the Great Lakes,
1825-1873. Part III. Inland Seas 8(2):99-104.

Odle, Thomas D. 1952c. The American Grain Trade of the Great Lakes,
1825-1873. Part IV. Inland Seas 8(3):l77-l92.

 

Odle, Thomas D. l952d. The American Grain Trade of the Great Lakes,
1825-1873. Part V. Inland Seas 8(4):248-252.

 

Odle, Thomas D. l953a. The American Grain Trade of the Great Lakes,
1825-1873. Part VI. Inland Seas 9(1):52-58.

 

Odle, Thomas D. l953b. The American Grain Trade of the Great Lakes,
1825-1873. Part VII. Inland Seas 9(2):105-109.

 

Odle, Thomas D. l953a. The American Grain Trade of the Great Lakes,
1825-1873. Part VIII. Inland Seas 9(3):l62-l68.

Odel, Thomas D. l953d. The American Grain Trade of the Great Lakes,
1825*1873. Part IX. Inland Seas 9(4):256-26l.

 

Parkins, Almon E. 1918. The Historical Geography of Detroit.
Michigan Historical Commission, Lansing.

Reid, J. and M. Schiffer, and J. Neff. 1975. Archaeological Consider-
ating of Intasite Sampling. Seupling iu_Archaeology, J. Muller
ed. 209-216, University of Arizona Press, Tucson.

 

Royal Grain Inquiry Commission. 1938. The Report by_the Royal Grain
Inquiry Commission, Ottawa.

 

 

 

Royce, Stephen. 1938. Geology of the Iron Ranges: The influence of
Geological Conditions on Mining Practice. In Lake Superior
Iron Ores, Chas Hickok ed. Lake Superior Iron Ore Assoc.,
Cleveland.

 

Schiffer, Michael B. 1972. Archaeological Context and Systemic Con-
text. American Antiquiuy 37:156-165.

 

Siegel, Sidney. 1956. Nonparametric Statistics for the Behavioral
Sciences. McGraw-Hill, New York.

 

Sokal, Robert R. and F. Names Rohlf. 1969. Biometry: The principles
and practice 2: statistics 12 biological research. W. H. Freeman
& Co., San Francisco.

 

 

 

Sommers, Lawrence M. 1977. Atlas o£_Michigan. Michigan State Uni-
versity Press, East Lansing.

Steel, Robert G. and James H. Torrie. 1960. Principles and Procedures
.2: Statistics. McGraw-Hill, New York.

 

255

Townsend, Sallie and Virginia Ericson. 1978. Boating Weather. How
52 Predict 25, What 52 22 About 15. David McKay Co., New York.

United States Coast Guard. 1970. Light List. U. S. Gov. Printing
Office, Washington, D.C.

Willey, Gordon and Jeremy Sabloff. 1974. ‘A History of American
Archaeology, Freeman & Co., San Francisco.

Williamson, Samuel H. 1977. The Growth of the Great Lakes as a
Major Transportation Resource, 1870-1911. Research i§_Economic
History 2:173-248.

 

Winkelman, A. 1971. Shipping Casulties Resulting £2 Total Loss gn_
the Great Lakes from 1879-1970.

Wolff, Julius F. 1969. Before the Days of Radar: Ship Collisions on
Lake Superior. Inland Seas 25(2):137-149.

Wolff, Julius F. 1979. The Shipwrecks of Lake Superior. Lake Superior
Marine Museum Assoc., Duluth.

 

Wright, Richard J. 1972. Inventory 2£_Shipwrecks within Michigan
Coastal Waters. MB on file Northwest Ohio-Great Lakes Research
Center, Bowling Green State University.

 

Zapfle, Carl and Earle E. Hunner. 1938. Preparation for Market
Requirements, Shipment and Reduction. In Lake Superior Iron
Ores, Chas. Hickok ed. Lake Superior Iron Ore Assoc., Cleveland.

 

"ITIElifl'flllfiflllfilifl'fl'flifll'TS

   

- -_-_-_fl_._-_-_-_-_-_-_‘_

A