v —- '- vw' u _

III

WW

II’

I

112
466

 

_THS_

SOUND - SOURCE LOCALIZATION
BY THE RED FQX

Thesis for the Degree of M. S,
NeiC‘rHtTJAE‘é STATE UNiVERSi’E’Y
THEWMXS EARL ISLEY
2969

THE/5'5

 

 

 

IHI 11 “MM!!!" 1111111111! 11111 (Ll ll \/

31

 

 

 

ABSTRACT

SOUND-SOURCE LOCALIZATION BY THE RED FOX

-.BY

Thomas Earl Isley

When prey animals must vocalize but must not reveal
their position, natural selection has favored nearly pure-
toned calls which are more difficult to locate than calls
of complex tones. In a controlled test I determined how
well nine red foxes located 13 different frequencies of
pure sound. Using food as a reward, the foxes were trained
to choose the correct location of a sound signal emitted
from one of two possible loudspeaker positions. The foxes
located 3.5 kHz better than all other frequencies and had
the most difficulty locating frequencies less than 900 Hz

and greater than 14 kHz.

SOUND-SOURCE LOCALIZATION BY THE RED FOX

BY

Thomas Earl Isley

A THESIS

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

MASTER OF SCIENCE

Department of Fisheries and Wildlife

1969

ACKNOWLEDGMENTS

I sincerely thank:

My major professor, Dr. Leslie Gysel, for suggestions
concerning experimental design, for encouragement through-
out the study, and for helpful criticisms on the text.

Committee members, Dr. John King and Dr. Rollin Baker,
for counseling on the design of the study and for helpful
criticisms on the text.

Mr. Charles Sokol and Mr. Richard Thomas for the use
of their electronic equipment.

The professors, fellow students, and others who
offered sincerely appreciated advice and encouragement.

I would like to express my deepest gratitude to my
wife, Marian, for her constant encouragement and for her

assistance in preparing the final text.

ii

TABLE OF CONTENTS

Page
ACKNOWLEDGMENTS . . . . . . . . . . . . . ii
LIST OF FIGURES . . . . . . . . . . . . . iv
INTRODUCTION . . . . . . . . . . . . . . 1
METHODS . O O O O O O O O O O O O O O I 3
Experimental Design. . . . . . . . . . . 3
Subjects . . . . . . . . . . . . . . 3
Procedure . . . . . . . . . . . . . . 3
Experimental Chamber . . . . . . . . . 3
Sound System . . . . . . . . . . . . 6
Calibration . . . . . . . . . . . . 7
Training and Testing . . . . . . . . . 7
RESULTS 0 O O O O O O I O O O O O O O O 10
DISCUSSION 0 O O O O O O O O O O O O O O 18
LITERATURE CITED. . . . . . . . . . . . . 23
APPENDICES
Appendix
A. Number of Correct Responses Out of 10 Trials
on Each of Two Replications. . . . . . . 25
B. Number of Correct Responses Out of 32
Trials of the Daily Control Treatment . . . 26
C. Total Number of Correct Responses for the Two
Replications. . . . . . . . . . . . 27

iii

LIST OF FIGURES

Figure Page
1. Test Apparatus . . . . . . . . . . . 5

2. The Mean Response in 20 Trials for Foxes One
Through Five. . . . . . . . . . . 13

3. The Mean Response in 20 Trials for Foxes Six
Through Nine 0 O O C O O O O O O O 15

4. The Mean Response in 10 Trials for All
Nine Foxes . . . . . . . . . . . 17

5. Calls Having the Characteristics for Being
Difficult to Locate . . . . . . . . 21

6. Calls Having the Characteristics for Being
Easy to Locate . . . . . . . . . . 21

iv

INTRODUCTION

Red foxes (Vulpes vulpes) respond to certain sound

 

signals in their predatory habits as indicated by their
approach behavior to artificial and recorded distress calls
of their prey (Morse and Balser, 1961; Busnel, 1963). The
acoustical detection used by foxes would be maximized if
all audible calls of the prey species could be readily
located; however, some vocalizations of the prey species
seem not to be useful in directing the fox toward their
sources. Certain birds signal the presence of a predator
by giving a continuous and nearly pure—toned call which is
difficult for humans to locate (Marler, 1955).

According to Marler and Hamilton (1967), vertebrates
locate sound sources by binaurally comparing differences in
intensity resulting from the sound shadow of the head,
differences in arrival time, and differences in phase.

They state that differences in arrival time are independent
of the frequency and are best determined if the sound con—
tains abrupt discontinuities, whereas detection of phase
differences requires a continuous sound. Locating a sound
by detecting differences in phase is possible for low

frequencies with wavelengths greater than twice the

distance between the ears, but at higher frequencies wave-
lengths are shorter and phase differences are difficult to
detect. At higher frequencies the head acts as a sound
shadow so that locating a sound by intensity differences
is possible only when the wavelengths are less than the
distance between ears.

Locating a sound by detecting differences in intensity
and differences in phase depends upon the head configuration
and may, because the two sound-locating mechanisms do not
overlap, result in an interval of sound frequencies more
difficult to locate than sounds of higher and lower fre-
quency. It would probably be especially difficult for a
predator to locate the source of a nearly pure-toned call
if it were pitched too high for locating by phase differ—
ences but yet not high enough for locating by intensity
differences. Predators, on the other hand, can presumably
best locate a particular call of nearly pure tonal quality
if their head configurations are such that detecting
differences in either phase or intensity is maximized at
that frequency.

This experiment was designed to determine how the
ability red foxes have for locating pure-toned sounds

varies as a function of the frequency.

METHODS

Experimental Design

 

The ability that a sample of nine foxes had for
locating 13 frequencies of pure sound ranging from
300 Hz to 34 kHz, was estimated in a controlled test situ—
ation. Also controls with no sound were tested to insure
that the foxes were locating by audition. The nine experi-
mental subjects and thirteen treatment levels were then
analyzed in a cross classified design with the number of

correct choices as the dependent variable.

Subjects

Five male and four female foxes were taken as pups
from four areas and raised in captivity at Michigan State
University where the experiment was conducted. The normal
diet consisted mainly of commercial dog food except during
testing when raw meat was used as a reward. The foxes

were tested at nine months of age.

Procedure

 

Experimental Chamber

 

The interior dimensions of the experimental chamber

were 1.8 m wide, 1.8 m high and 3.6 m long (Figure l).

.msumummmm ummBII.H musmflm

 

T E _ 1

 

 

  

 

 

 

 

 

 

 

 

  

 

 

  

 

 

23.550 15.040 7/
It; mozwu‘ W
// ozgmm /
moon \ // /rmooo :51 $8
.. 20:50.. on. .N/\ / // 823:5
4.” $3.4QO meg @2525 ~98
. _ mwxom mozmm %oz_ouox
382;, /co/oou_ 62.92%
WW 2063 «000 txmi.
>43 uzo \\ z
\u /
/
w
, gig;

 

 

The door, located at one end of the chamber, had a 15 by

30 cm one-way vision window horizontally mounted 1.5 m
above the floor. The walls and ceiling of the chamber were
carpeted on both sides and egg flats were fastened on the
inside of the walls and ceiling to suppress echoes. The
floor of the chamber consisted of a 5 cm layer of fiber-
board and three layers of carpeting. The steady-state
ambient noise level in the chamber averaged 53 t 3 db (re
0.0002 dynes/cmz) when measured on the linear scale of a
Brfiel and Kjaer type 1613 sound level meter. The room was

illuminated by a 300 watt incandescent light bulb.

Sound System

 

Thirteen test frequencies (0.3, 0.6, 0.9, 2.0, 3.5,
5.0, 6.5, 8.5, 10, 14, 18, 26, and 34 kHz) were generated
by a Hewlett-Packard model 0—10/APA—6X audio oscillator and
fed into one of two transducers. The accuracy of the fre-
quencies generated were checked with a Tektronix 502A dual
beam oscilloscope. A University model T-202 sphericon
super tweeter was used for the frequencies 6.5 kHz and
higher and a JFD model ALC-l speaker was used for the
frequencies 10 kHz and lower. This allowed a comparison
of the two speakers at 6.5, 8.5, and 10 kHz. The fre-
quencies used were all within the range of hearing for red

foxes (Peterson, 1969).

Calibration

 

The system was calibrated with the B and K sound
level meter located at various positions in the starting
cage. The intensity of all but the highest two test tones
and from both speaker locations was kept at 74 t 4 db (re
0.0002 dynes/cmz) when measured on the linear scale. The
highest two frequencies were calibrated with the frequency
response curve supplied by the manufacturer of the Uni-
versity speaker. A VO Matic 360 volt meter was used to
insure the voltage output of the signal generator remained

constant.

Training and Testing

 

The foxes were trained to choose one of two possible
positions from which the sound source was coming then to
approach and open the door of a food reward box located
below and in front of the loudspeaker. A correct choice
was rewarded with a 10 gram piece of raw meat. The appara—
tus, similar to the one used by Neff gt 31. (1956), con—
sisted of an intertrial holding cage, located outside of
the chamber, a starting cage where the animal received the
sound stimulus, a 1.22 m high separating fence, and a cloth
curtain attached to a fence behind the two food boxes
(Figure 1). The food boxes were 15 cm wide, 12 cm deep
and 22 cm high with 11 by 14 cm hinged doors and were Open
in the back to facilitate placement of the meat. The

speakers were centered 8 cm above the food boxes and were

positioned behind the cloth curtain to avoid visual cues.
The speakers formed a 35.5 degree test angle with the
starting cage as the apex.

At the beginning a preliminary control test was
conducted to determine if the animals could locate the
meat by odor alone. With the food boxes one m from the
starting cage and a 58 degree test angle, the foxes could
not locate the meat.

Training started at three months of age by condition-
ing the foxes to associate food with the sound stimulus.
At five months of age the foxes were consistently respond-
ing in the test apparatus. Pilot studies then indicated
that a test angle of 35.5 degrees elicited a less than per—
fect response from the foxes.

A trial consisted of putting the speaker and food
in place while a fox waited in the holding cage, turning
on the stimulus, letting the fox into the starting cage,
five seconds later releasing the fox from the starting
cage, and turning the stimulus off after the fox had
reached the food box and eaten the meat or after he made
an error. The dividing fence extended above the starting
cage and prevented the animals from reversing their choice.
The appropriate exit door was opened and the foxes were
trained to exit and wait in the holding cage until the next
trial. Speaker location was determined by the Gellermann
series (Gellermann, 1933) which is different combinations

of left-right choices. During the speaker and meat

placement all movements were kept constant to avoid giving
the animals additional cues. Intertrial time was kept a
constant one minute.

When the actual testing began, each animal received
10 trials daily of a randomly determined frequency until
all 13 frequencies and a control of no sound had been
tested twice. The control was identical to the other
treatments except the sound stimulus was not turned on.
The frequencies 6.5, 8.5, and 10 kHz were tested twice
with each speaker for comparison between speakers.

One final control was used to insure that the animals
were not locating the sound while they were yet in the
holding cage. Each day after testing every fox, an addi—
tional trial was given where the sound stimulus was turned
on and off before the fox entered the starting cage. The

rest of the trial was identical to the regular trials.

RESULTS

The data for the entire testing period are presented
in Appendices A and B. For the statistical analyses a
response is defined as the correctness of choice made by
a fox.

x2 goodness of fit tests indicate the foxes responded
randomly to both the daily control trials (P > 0.70) and to
the replicated control test (P > 0.80). Therefore, the
test situation allows making the assumption that a better
than random response may be attributed to the foxes locat-
ing the sound source.

The speaker used at 6.5 kHz was randomly determined
for each fox since a Mann Whitney U test shows that the
foxes did not differ in their response to the two speakers
at 6.5 kHz (P > 0.10). The University speaker was used for
all frequencies higher than 6.5 kHz because at 8.5 and 10
kHz the foxes responded significantly more accurately to
it than to the JFD speaker (P < 0.05). The data used for
the final analyses of treatment and animal differences
appear in Appendix C.

There are highly significant differences between

frequencies (P < 0.001) but the foxes do not differ

10

11

significantly (P > 0.10). A Friedman two-way analysis of
variance was used to test for frequency and animal differ-
ences because the data are skewed toward complete accuracy
in response. The mean response of each fox at all of the
13 frequencies is given in Figures 2 and 3 and the mean
response of all foxes at all of the 13 frequencies is given
in Figure 3.

The following trends are apparent from Figure 4.
Starting at 0.3 kHz the foxes show a gradual increase in
their response until 3.5 kHz where all foxes reach a peak
response. As the frequency increases from 3.5 kHz, the
foxes show an average decline in response with slight dips
at 8.5 kHz and 18 kHz. Mann Whitney U tests show that the
foxes differed significantly in their response at 8.5 kHz
and 18 kHz when compared to the next higher frequencies
respectively (P < 0.05).

Figures 2 and 3 show that the foxes were not identical
in their response. Some of the fox X frequency inter—
actions might be attributed to one poor test out of the
two tests for each frequency. Examples of this are fox
numbers four and six when tested at 2.0 kHz. Fox number
four got 5 out of 10 and 10 out of 10 correct responses
during the two tests at 2.0 kHz. Fox number six got 5
out of 10 and nine out of 10 correct responses during the
same two tests. Foxes seven and nine show a definite
increase in response at 34 kHz whereas the rest of the

foxes remain unchanged or fall off in their response.

12

Figure 2.--The mean response in 20 trials for
foxes one through five.

NUMBER OF CORRECT CHOICES

13

 

 

 

 

 

Fox No. Id" _.__._.
g \,_,_,,,_, ,
6 /\.
4 A
2 1 l4 l1l11| 1 111111111 1 1 111111111]
0 Fox No.2 ._ _.
8 90/ \/ \ .‘o
/
6 /
4 o
2 1 l 111l11l 1 11111111] 1 l 1111J1l11J
Fox No.3 C? o
o ./.’ \.\.\/.\
8 g / O O
6 \. \/\'
4 o
2, 1 l lJJllll 1 11111L11l 1 1 111111111]
Fox No.4 ,
o Q .\/.\. \’\.\o—
. \
6 .x. O
4
2 J 111L141] 1 111 llLLll 1 11111I1I1IJ
0 FOX N0. 5 9 ._._K. g
8 O#.\./ . \.’.
6 \
4
20' 1 1 10151111] 1 1L111J11l 1 l l11l1l11J
LO 5.0 LC 50 100

FREQUENCY (kHz)

14

Figure 3.--The mean response in 20 trials for
foxes six through nine.

NUMBER OF CORRECT CHOICES

15

N‘hmmo

1 1 1111111 1 11111111114 1 11111111

Fox No.7 q/.\/.-.q\/\. .

E o
E / \ ‘/
' /
1 1 1111111 1 1111111111 1 1 1 1111L1J

N¢O>QO

4|
4)

Fox No.8 9 .1 /.
././ \V \\.’,

1 11111111 1 1111111111 1 4 11111111

memo

Fox No.9 (7

/"""’"" '\~-~~«~/\/

7 1 1 1111111 1 J11 1111111 1 1 111111141
.| 0.5 |.O 5.0 IO

FREQUENCY ( kHz 1

l6

.mmx0m mafia Ham now mamfluu OH a.“ omcommou some 93.11;. ousmam

l7

Aux: >ozm30mmm

 

 

 

 

0” ON 0. OF 0.0 0.m 0.N 0.. 50 0.0 ”.0 N.0 _.0
- — _ q d u q _ a — q fid - q u — q a —
va
.
. Irk-
.
[rm
I‘m
.11. .\./ \1 11.
. LVO—

SEQ/OHO 1938800 :10 HERA/RN

DISCUSSION

The foxes responded best between 0.9 and 14 kHz with
a high response at 3.5 kHz and a low response at 8.5 kHz.
If frequencies lower than 8.5 kHz are located by phase
difference detection and frequencies higher than 8.5 kHz
are located by intensity difference detection, then the
low response at 8.5 kHz could be attributed to the two
locating mechanisms not overlapping. The tympanic mem-
branes in adult red foxes are approximately four cm apart
so phase differences should become difficult to detect
above 4.2 kHz and intensity differences should be difficult
to detect below 8.4 kHz.

Osterholm (1964) reported that a single fox located
700 Hz more accurately than all higher frequencies. The
reason his results differ from the results of this study
may be because he used a different test situation and
because his fox was tested at a younger age.

The pinnae might enable foxes to binaurally detect
phase differences at frequencies higher than those pre-
dicted by the theory. Batteau (1967) contends that the
human pinna has a significant role in auditory locali-

zation by temporarily delaying the incoming auditory signal

18

19

as a function of the pinna position relative to the sound
source.

An animal must emit sounds which provide a minimum
of locating cues if revealing its position is disadvan—
tageous. This is done by avoiding abrupt discontinuities,
hence minimizing binaural detection of differences in time
of arrival, and also by using a nearly pure tone minimizing
binaural detection of differences in phase and intensity.
Examples of such calls are the flock call of ring-necked

pheasant chicks (Phasianus colchicus) and the wheeeep

 

call (Sheldon, 1967) of a lost woodcock chick (Philohela
miner) (Figure 5). The flock call is emitted by pheasant
chicks under stress and is difficult for humans to locate
(Heinz, personal communication). The wheeeep call of wood-
cock chicks is also difficult for humans to locate (G. A.
Ammann, personal communication).

Human beings have more difficulty in locating the
frequencies of sound between 2.0 and 5.0 kHz than the
immediate higher and lower frequencies (Stevens and Davis,
1938; Marler, 1955). These frequencies are too high to
locate by binaurally detecting phase differences and too
low to binaurally detect intensity differences. Although
a call can be structured to offer a minimum of locating
cues, humans and foxes probably do not have the same
ability for locating the call. The juvenile woodcock
wheeeep call and the juvenile pheasant flock call are

examples of calls that foxes might locate better than human

20

Figure 5.--Calls having the characteristics for
being difficult to locate. (Spectrograph of flock call
after Heinz and Gysel, in press.)

Figure 6.-—Calls having the characteristics for
being easy to locate. (Spectrographs after Heinz and
Gysel, in press.)

1 KILOHER‘I’Z )

FREQUENCY

FREQUENCY 1 KIIOHERTZ 1

 

FLOCK CALI.

21

WHEEEEP CALI.

 

 

 

 

o- l

5 u I

2 - l
I r I I U I I l r E i 1— ' —I
o 2 4 o a to 12 o 2 4 o a 10

TIME (tmms or A SECOND)

11- noon «mama CALI. I

o - I

‘1 1 11 ‘ ‘11 (\I I

2 ‘ I I a I I l “WM 1- '1

1 my 1110*

5 3 1 2 3 1'o 1'2 3 Z Z 2

TIME (TENN-IS OF A SECOND)

22

beings. Both of these calls are pitched around the fre-
quencies that foxes appear to locate the best and humans
experience some difficulty in locating.

To be readily located, a vocalization should contain
a wide range of frequencies and abrupt discontinuities
providing for binaural detection of differences in both
phase and intensity as well as detection of binaural time
differences. The brood gathering call of the hen and the
crowing call of the male ring-necked pheasant are examples
of vocalizations having these characteristics and are
shown in Figure 6 for comparison to the calls of Figure 5.

Future research should determine which sound signals,
and their properties, red foxes respond to in their preda-
tory habits. Many of the prey species taken by red foxes
use sound signals that are pitched within the frequencies
that foxes readily locate; however, the young of some
rodents can emit high frequency calls, ranging from 17 to
80 kHz, which would be difficult for foxes to locate (Hart
and King, 1966; Noirot, 1966; Noirot, 1968; and Noirot and
Pye, 1969). It would be interesting to determine which
property of a sound signal elicits the response from a fox.
Finally, it is yet to be determined how much more readily
foxes locate discontinuous sounds. of varying intensity and
frequency, than pure-toned sounds. Such a study could be
done in this type of test apparatus using different combi-
nations of frequency, intensity, and discontinuities in the

sound.

LITERATURE CITED

LITERATURE CITED

Batteau, D. W. 1967. The role of the pinna in human
localization. Proc. Roy. Soc. 168:158-180.

Busnel, R. G. 1963. Aspects of animal acoustic signals.
In R. G. Busnel (ed.) The acoustic behavior of
animals. Elsevier Publishing Company, New York.
933 p.

Gellermann, L. W. 1933. Chance order of alternating
stimuli in visual discrimination experiments.
J. Genetic Psychol. 42:206-208.

Hart, F. M. and J. A. King. 1966. Distress vocalizations
of young in two sub-species of Peromyscus. J. Mammal.
47:287-293.

Heinz, G. H. and L. W. Gysel. In Press. Vocalization
behavior of the ring-necked pheasant. Auk.

Marler, P. 1955. Characteristics of some animal calls.
Nature 176:6-7.

Marler, P. and W. J. Hamilton. 1967. Mechanisms of animal
behavior. John Wiley & Sons, Inc., New York. 771 p.

Morse, M. A. and D. S. Balser. 1961. Fox calling as a
hunting technique. J. Wildl. Mgmt. 25:148-154.

Neff, W. D., J. F. Fisher, I. T. Diamond and M. Yela. 1956.
Role of auditory cortex in discrimination requiring
localization of sound in space. J. of Neurophysi-
ology 19:500-512.

Noirot, E. 1966. Ultra-sounds in young rodents. I.
Changes with age in albino mice. Anim. Behav.
14:459-462.

Noirot, E. 1968. Ultra-sounds in young rodents. II.

Changes with age in albino rats. Anim. Behav.
16:129-134.

23

24

Noirot, E. and D. Pye. 1969. Sound analysis of ultra-
sonic distress calls of mouse pups as a function of
their age. Anim. Behav. 17:340-349.

asterholm, H. 1964. The significance of distance receptors
in the feeding behavior of the fox, Vulpes vulpes.
L. Acta 2001. Fennica 106:1—31.

 

 

Peterson, E. A., W. C. Heaton and S. D. Wruble. 1969.
Levels of auditory response in fissiped carnivores.

Sheldon, G. 1967. The book of the American woodcock.
The University of Massachusetts Press, Amherst.
227 p.

Stevens, S. S. and H. Davis. 1938. Hearing. John Wiley
& Sons, Inc., New York. 489 p.

 

APPENDICES

25

 

 

 

 

uoxmmmm >UHmum>HcD .lwAn uwxmmmm own 1w

A

 

 

m OH oH O OH O m O O O OH OH oH OH OH O m m

m OH O O OH OH O m h O OH O OH OH O b O

m m h O OH O m m m h m O OH OH OH O h O

O h m O OH OH O m m h O OH OH OH OH OH h

h O OH O OH OH O OH OH 5 OH OH OH OH OH O m h

O OH O m O OH O OH O O OH OH OH O OH OH O

m O O O m b O O OH O m O OH O OH O O O

O h m h m h m m h 5 OH 5 OH m OH O h

m h m O OH OH O m O O O OH OH OH O m m O

O m m O OH OH O OH OH O OH OH OH OH O m m

O O m m O OH OH OH O O OH OH OH OH OH 5 O Q

m O . m O. OH OH O OH O OH OH O OH O OH O O

O O m O O OH O OH O O OH OH OH OH OH O m m

O O h e O OH 5 m O O m O OH O h m h

m h m O OH OH O OH h O OH OH OH OH OH OH O N

m m O O O OH OH OH OH O m s OH OH O O v

m v w O OH O O OH OH O OH O OH OH OH OH O H

m h m m m m m O OH O OH O OH OH OH OH O
Houu vm ON OH OH OH m.m m.O OH m.m m.O O.m m.m O.N m.O 0.0 m.O
1:00 xom

Anmxv mocmsvmum

 

H.60Huoa mcHumwu muHucm may ucommumou sumo mmmnaa
.mcoHumoHHmmu 03» mo sumo co mHMHHu OH no use mmmcommou uomuuou mo H02E9211.< xHazmmmm

26

APPENDIX B.--Number of correct responses out of 32 trials
of the daily control treatment.

 

 

 

£23235.

1 19

2 17 j
3 15 5—

4 14

5 23

6 15

7 18

8 . 16

9 13

 

27

 

 

T11

I
‘

uoxmmmm OuHmHo>HCD A. umxmmmm GOO .w

 

 

ONH OOH OHH OOH OOH HOH OOH OOH OOH OOH OOH OOH OHH mHmuOB

ON OH OH ON OH OH OH OH ON ON OH OH HH O
OH OH OH ON .OH OH OH OH ON ON ON OH OH O
OH OH OH OH ON OH ON ON ON OH ON OH OH O
NH OH OH OH OH OH OH OH ON OH ON OH OH O
OH OH OH ON ON OH OH ON ON ON OH OH OH O
NH OH OH OH ON OH ON OH ON OH ON HH NH O
NH OH O OH ON OH OH OH ON OH OH HH OH O
OH OH OH OH OH OH OH OH ON ON OH OH O N
HH OH O OH OH OH OH OH ON ON ON ON OH H

 

Om ON OH OH OH 0.0 0.0 0.0 O.m O.N 0.0 0.0 m.O

 

xom
Aumxv mocoavoum

 

H.mmocmuwmeO HmEHcm can wocosvmum mo mmmOHmcu Hmch on» :H Owns mums
mumw mmmsea .mcoHumoHHmou o3u on» How momcommmu uomuuoo mo Hogans Hmuoa11.o xHozmmmd

MICHIGAN STATE UNIV. LIBRARIES

”Hill
31

1

"I‘ll!" Ill" 1111»!
9

2 31002352

1
1