Studies on Amerindian dogs, 2: Variation in early peruvian dogs

Journal

Science 1979,6, 139-161

of Archaeological

Studies on Amerindian Dogs, 2: Variation in Early Peruvian Dogs Don Brothwell,ll A. Malagaband Richard Burleigh” Remains of dogs have been discovered in Peru for over a hundred years, but relatively little information is available on them. Dating evidence, especially by the radiocarbon method, suggests that the skulls of dogs considered in detail here range from 1030 BC to 1324 AD, with the majority belonging to the past two millennia. Craniometric information is presented on a series of early Peruvian specimensnow preserved in Lima and London, representing the majority of material as yet unpublished, and the largest sample so far studied. Additional Peruvian data were also collected from the literature, as well as comparable information on some other Amerindian dog series. It has been possible to indicate that two distinct forms are represented in the Peruvian material. AMERICA, ANCON, BONE, CANID, -CARBON-14, CRANIOMETRIC VARIATION, DOG, FOOD, HAIR, NEW-WORLD, PENROSE STATISTIC, PERU, SEXUAL DIMORPHISM.

Keywords:

Purpose and extent of present study We have restricted ourselves in this study to a number of aspects concerned with early Peruvian dogs. First of all, it seemed essential to find out exactly how much canid

material has been excavated and to investigate at first hand as much of this material as possible. The dating of the specimens was clearly an important step, especially as some specimens were poorly dated by cultural associations. The work was initiated by one of us (D.R.B.) on the sample of dogs from Ancon (Figures llr), that was acquired by the British Museum (Natural History) by H. 0. Forbes in 1913, and as yet unstudied. These data were then combined with the significant collection in the Laboratory of

Palaeoethnozoology in Lima curated by one of us (A.M.) and derived from various sites. Further Peruvian dog material was added, and although we do not claim to have seen all canid skeletal remains now available from Peru, we feel our data warrant this initial survey. As regards the osteometric information, we plan to restrict ourselves to the craniometric data. Odontometric variation is still in the process of study and will

be published separately.

03054lO3/79/020139+

‘Institute of Archaeology, 31-34 Gordon Square, London WCl, England. bLaboratorio de Paleoetnozoologia, Instituto Veterinario de Investigaciones Tropicales y de Altura, Universidad National Mayor de San Marcos, Apartado 4480, Lima, Peru. cResearch Laboratory, The British Museum, London WCl, England. 139 24 $02.00/O @ 1979 Academic Press Inc. (London) Limited

140

D. BROTHWELL

ET AL.

STUDIES

ON AMERINDIAN

DOGS,

2

141

142

D. BROTHWELL

ET AL.

Figure 3. Lateral views of early Peruvian dog skulls now in the University collection in Lima, (a) A168; (b) A165; (c)Al62; (d)A161 ;(e) A164; (f) AIM.

Finally, in relation to the osteometric data, we feel it is time to debate the variation seen in Peruvian dogs in relation to possible regional differences, the likelihood of heterogeneity, and also the possibility of early dog breeding (albeit at a low level). Bones found on many archaeological sites are the result of human exploitation of available resources of animal food. Although Amerindian dogs were used as food, the majority studied here were presumably animals killed for ritual purposes, as skin and hair were still intact. On available archaeological evidence, it is not clear how far there may have been a biassed selection for cultural reasons. Material available for craniometric study As much skull material as possible has been measured, with a view to forming a core of data against which further Peruvian and other Amerindian dog material can be compared. In a number of instances the skulls are damaged, or dried soft tissue and hair partially cover the bones, so that complete runs of measurements are uncommon. In view of the smallness of the sample, our means are calculated on measurements from both males and females, only dogs with erupted and functional second molars being included. Measurement numbers are those given by von den Driesch (1976), and we have endeavoured to follow her definitions precisely. As we wish to concentrate on intra-group variation and not extend for the moment on to the larger task of intergroup differences in Amerindian dogs, statistical treatment is minimal but we feel sufficient. Individual cranial measurements are given in Tables 1 and 2.

STUDIES ON AMERINDIAN

DOGS, 2

143

Figure 4. Lateral views of early Peruvian dog skulls now in the National Museum of Anthropology and Archaeology in Lima. (a) Paracas S/N2; (b) Ancon Al/10750 d; (c) Ancon Al/10750 a; (d) Ancon Al/10750 b; (e) Paracas C1; (f) Ancon Al/10750 c.

A. British Museum (Natural History). Dogs from Ancon (1) 729 Parts of an immature skeleton. Skull collapsed. Deciduous dentition (2)

95

(3)

243

(4)

251

(5) (6) (7)

250 635 DB(a)

(8)

DB(b)

B.

only. Cranium with damage to right zygomatic area. Adult. Skull partly covered by flesh. Also the tail region. Permanent teeth unworn. Skull. In good condition and partly covered in dried soft tissue and hair. Adult. Complete adult skull mainly covered in flesh. Well worn teeth. Much of the skeleton, including a well preserved skull. Adult. Much of the skull of an adult animal, but with post-mortem facial damage. The broken skull of an immature dog of about 4-5 months.

National Museum of Antiquities, Lima. Dogs from four sites

(1) (2) (3) (4) (5) (6)

Ancon. Ancon. Ancon. Ancon. Paracas. Paracas.

Excavated by J. Tello. Al/10750 a. Broken and incomplete cranium. Excavated by J. Tello. Al/10750 b. Well preserved cranium. Excavated by J. Tello. Al/10750 c. Cranium in good condition. Excavated by J. Tello. Al/10750 d. Incomplete skull of sub-adult. Reasonably well preserved cranium. Cl. S/N2. A sub-adult incomplete skull.

1 2 3 4 5 1 10 12 15 17 18 19 20

119.4 120.4 113.9 104.5 99.0 68.8 33-o 30.0 20.8 28.3 48.1 21.8 17.9

Al63(L)

15.9 22.5 40.1 18.7 14.8

961 91.9 93.1 80.8 78.0 -

A166(L)

102.6 103.2 99.4 88.2 85.6 60.8 30.5 27.2 16.8 244 42.8 19.5 15.6

4154(R)

Table

121.3 121.6 115.9 105.8 100.8 69.0 32.8 31.1 20.7 28.9 51.3 21.9 18.3

from

108.8 110.7 105.4 93.0 90.0 63.3 30.7 21.4 17.5 26.2 46.1 18.9 16.7

A162(R)

mandibles

A155(L)

2. Dog

112.6 112.5 109.2 94.1 91.8 64.2 32.4 29.6 18.8 26.8 45.0 19.1 17.6

Ancon 250 112.4 121.4 115.7 106.7 100.2 70.0 32.3 30.3 21.7 29.7 51.9 19.2 19.4

113.6 115.5 109.7 91.5 94.3 -_ 30.8 19.5 25.7 50.1 16.4 16.2

as defined

Ancon 635

(measurements

Ancon 243

Peru

19.4

53.2

129.2 129.5 125.4 112.3 107.6 72.1 32.9 32.3 20.2

Ancon 251 126.8 128.2 123.6 109.2 106.2 73.2 35.0 32.5 21.0 32.0 52.6 21.7 19.2

Ancon DBY

in van den Driesch,

19.8 28.6 47.8 18.4 16.4

34.1

-

-

Ancon DBx

1Y76)

13.7 33.9 31.2 23.0 31.8 51.4 21.0? 18.4

-

Ancon DBz

110.4 I1 1.5 106.2 97.3 93.2 IO.5 33.4 29.0 17.7 23.4 42.1 17.0 15.9

Cajamarca

1

STUDIES ON AMERINDIAN

(7) (8)

DOGS, 2

14.5

Huacho (Chancay Epoch). Cranium in reasonable state of preservation. Cerro Sechin. Incomplete skull of an immature dog.

C. Laboratory of Palaeoethnozoology,

(1) (2) (3) (4)

Al54 Al55 Al57 Al58

(5) (6) (7)

Al62 Al63 Al67

University of San Marcos, Lima

Pando. Young adult skull showing minor upper P3 rotation. Pando. Skull in generally good condition. Possibly a young adult. An immature but well preserved skull. Canines not fully erupted. An immature skull with deciduous dentition in position. Nearly 3 months. Chancay. Young adult skull in good condition. Huaycan. Mature skull. Considerable tooth wear. Upper P3 rotated. Puerto Viejo. Fragmentary skull of an immature dog.

Dogs from Mala (Seaside Groves, Nos 8-16) (8) Al59 Incomplete skull of an immature dog. Damaged cranium of a sub-adult dog. (9) A160 (10) A161 Young adult skull in generally good condition. (11) Al64 Cranium of a young adult. (12) Al65 Incomplete cranium, lacking zygomatic areas. Adult. (13) Al66 Incomplete cranium showing rotated upper P3’s. Broken cranium of an adult. Rotation of left upper P3. (14) Al68 Partial cranium. Adult. “Bulldog” nasal area. (15) A170 Damaged and incomplete cranium. (16) A171 D.

Escuela Politecnica National, Quito, Ecuador

(1)

Peruvian skeleton from Cerro la Vaqueria, Cajamarca. Perfect condition. Adult. Sexual dimorphism in the Peruvian dogs

Although one might expect noticeable sexual dimorphism in dogs in view of its occurrence in some carnivores (Kurt&, 1955), there is in fact a singular lack of information concerned with this possible variation. In particular, one might have expected differences in the development of the temporal line, sagittal crest, the facial region, size of teeth, and perhaps robustness in the nuchal area of the occipital. However, the only region which seems to indicate the sex of the dog with any degree of reliability is the basilar part of the occipital bone. The & Trouth (1976) found in their study of 80 dogs of known sex that 87.5% of their skulls could be identified correctly. In view of this, and the relevance of sexing not only in osteometric work but also for the cultural inferences (e.g., whether there is a bias related to the ritual burial of some Amerindian dogs), we have attempted to apply this method of sexing to the Peruvian specimens. Postmortem damage or extreme immaturity were factors which resulted in the exclusion of a number of specimens. For the present study, then, we have attempted to apply the findings of The & Trouth (1976), evaluating the Peruvian basi-occipitals according to whether there was a broad or triangular-like elevation around the tuberculum pharyngicum (the female having a broader pattern). This may seem variation of a questionable sexual dimorphic character, but The & Trouth reasonably point out that the modelling of this basal area is associated with the attachment and degree of development of the rectus capitis ventralis major and minor muscles, the medial attachments of their fasciae on both sides demarcating a triangular-like area around the tuberculum pharyngicum. It would appear that these muscles are better developed in the male and thus their attachment area is deeper

146

D. BROTHWELL

ET AL.

5cm

Figure 5. Sexual dimorphism, as seen in the cranial base of four early Peruvian skulls. Males are (a) Huacho; (b) Paracas C,; (d) Al/10750 a. The female is (c) Al/10750 c.

Table

3. Tentative

estimates

of sex in the Peruvian

Huacho (Lima) Paracas C 1 BMNH 243

DB (4

Al/10750 a Al/10750 b A 168 Al/10750 c BMNH 251 A 161 BMNH 95

4

3

4

dogs

Al/10750 d A 160 A 162 BMNH 250 A 161 5

and more medially extended. The resulting variation can be illustrated by reference to early Peruvian cases (Figure 5). The results would seem to indicate that this basal dimorphism can be expected and thus usefully employed in early dog samples. As the muscles are called into play during head flexion and male fighting, one might reasonably speculate that differences may now have evolved in different breeds related to reduced musculature and selection for less aggressive animals. In the Peruvian dogs, those which could be tentatively sexed were assigned as 6, $?, $2or o?. The proportion of dogs in each of these groups was about the same (Table 3), and there is no indication in this small sample of any change through time, or by comparing the Ancon sample with the rest. It seems, then, that the sex of the dog was not important in the ritual use of dogs in early Peru. The possibility that the Peruvians could have intentionally related male dogs with male human burials, or females with females, cannot at present be resolved owing to the paucity of associated archaeological information.

STUDIES

ON AMERINDIAN

DOGS,

2

147

Earlier studies on Amerindian dog biology Much of the early literature on biological variation in Amerindian dogs was reviewed by Allen (1920) and does not warrant extensive discussion again here. The exception is the work by Alfred Nehring who, between 1884 and 1887, published various notes on early Peruvian dogs, in particular those derived from excavations at Ancon. Although the number of dogs available to Nehring was small, he considered that the variation was sufficient to separate the material into three different varieties, namely: a “turnspitlike Inca dog” (classified as Canis ingae veutagus); secondly, a “collie-like Inca dog” (Canis ingae pecuavius); thirdly, a “bulldog or pug-like Inca dog” named Canis ingae molossoides.

By the time of Allen’s (1920) classic paper there was clearly some reluctance to go so far as to give species and subspecific names to the variation seen, whether it was on the evidence of coat colour, coat length, skull shape, limb proportions, or a combination of these. Instead, Allen considers seventeen major “breeds” of Amerindian dogs, although he points out that there are really two chief forms of dog, a noticeably large and a small variety. Nevertheless, on all the New World evidence available to him, he concluded that there were “apparently sundry local breeds . . . probably conforming in distribution with the general areas occupied by the groups of tribes amongst which they were found”. He realized that skeletal traits might “frequently fail to give any clue to external traits that would be distinctive”. Furthermore, he saw the problem of using the term “breed” in his studies, when its application to local variation in Amerindian dog groups was not the result of “any conscious effort to change or keep constant the traits” by the human groups. As regards early Peruvian dogs, Allen (1920) was clearly impressed by the long-haired form of so-called Inca dog, and accepted this group as distinctive. The hairless dog in South America he also sees as a variant with a distribution from Central to South America. The separation of groups of dogs on the basis of size alone is of course especially fraught with danger, but he identifies a “small Indian dog” in Peruvian material from Chicama and Coyungo. Nasal length also shows recognizable variation, and he accepts the occurrence of a “short-nosed” dog at Ancon. Evolving from this form, he suggests, is the “Peruvian pug-nosed dog”, also found at Ancon, as well as Pachacamac. The distinctive features of this cranial form are given as: (a) very short face, (b) undershot lower jaw, (c) broad zygomatic arch, (d) broad nasal passage, (e) an upward turning and distinctive angulation of the palate. Recognition of distinctive forms, on skeletal evidence, continued over the next few decades, although there is some further confusion caused by the employment of modern breed names. Ritchie (1945) for instance, says that “two breeds of dogs” are represented at the Frontenac Island site, one “perhaps terrier-like” and the other “resembling in size and build a collie or shepherd dog”. A report on early New Mexico dogs by Allen (1954) substantiates, it is suggested, the occurrence of a “Plains Indian Dog”, both on length of hair and cranial features. Although concerned with North American dog material, Haag (1948) produced the first extensive metrical study of aboriginal dogs, and in so doing set a new pattern for future investigations (not often lived up to). In his study of a desiccated dog from Puerto de Supe, Peru, Haag (1954) would seem to accept the occurrence of a medium-sized, short-faced form, but does not believe it could fall into Allen’s category of “Peruvian pug-nosed”. In apparent contrast to his comments on variation in some groups of dogs, he notes a marked uniformity in the early dogs along the Pacific coast (Haag & Heizer, 1953). Similarly, Stewart (1963) undertook a metrical study of late prehistoric dogs from Delaware, and also concludes that the variation does not justify the recognition of more than one homogeneous population.

148

D. BROTHWELL

ET AL.

A little over a decade ago, “breed” was still surprisingly being used as a worthwhile term. Barbara Lawrence, who has made a significant contribution to the study of early dogs, is content to say that there was a tendency, in some areas, “for dogs to form local. fairly well-defined breeds” (Lawrence, 1967). In a subsequent study of early North American dogs Lawrence (I 968) suggests that both large and small forms of dog have some considerable antiquity in the New World. Olsen (1974) suggests that at least “four size categories” should be recognized in North America alone. He points out (Olsen, 1976) that following European contact, further variation in size and shape was intruded into the Americas. It could be, of course, that this was the first appearance of true breeds, rather than the occurrence of regional (racial/subspecific) variation which was the result of micro-evolutionary processes and not intentional human selection. This does not mean to say that dogs have not been kept and valued for specific purposes. Wing (1975) rightly points out that Andean dogs have been kept for human protection and to defend flocks. They were also used to follow trails and catch birds. Dog meat was eaten by some human groups in South and Central America, as well as at times also by more northern communities (White, 1955). In the case of the numerous dog burials which have been found at Ancon, it seems likely that they were regarded as having special ritual value, and were of assistance to the dead, in their passage to the next world. In the case of the hairless Peruvian dog, this has been considered in detail by Weiss (1976), who concludes that it is an hereditary condition with a history of some 2000 years. This is not a “race” or “breed” but a distinct teratogenic variety which behaves as a Mendelian dominant, affecting the skin, teeth and claws. Weiss considers that no mummified or skeletal material of this dog anomaly has so far been found. Although apparently displaying a short face, Weiss considers it to be separate from the so-called “bulldog” form (because of its dental abnormalities, etc.). The literature on New World dogs, including Peruvian specimens, is thus not without comment on variation. indeed, the hundred years of discovery and study which have occurred since Nehring’s early work, has resulted in various schemes for the classification of Amerindian dogs into a number of groups. It is now our intention to look into this possibility from the point of view of cranial variation in Peruvian dogs. Craniometric Variation The individual measurements of the skull in the more complete adult and sub-adult Peruvian dogs studied are given in Tables 1 and 2. Measurements are as defined by von den Driesch (1976), and the measurement numbers are those given by her. Comparison of these data with other Amerindian material has proved to be extremely difficult. The reasons may be summarized as follows: (a) Detailed studies of dog samples are limited, and are mainly concerned with the more eastern part of the USA. (b) The publication of basic craniometric data or means and other statistical parameters leaves much to be desired. In few series were more than six or seven measurements compatible with those given in von den Driesch. Thus, although extensive lists of measurements are given for various geographic racesj subspecies of American wolf by Young & Goldman (1964), only seven of their dimensions had comparative value in relation to this present study. (c) Although Giles (1960) and Lawrence (1967) have demonstrated the value of applying multivariate statistics to questions of coyote and wolf variation, the extension of this type of work to dog samples has been surprisingly lacking, except for one or two exceptions concerned with early European dogs. The recent application of numerical methods to problems of canid taxonomy in general

STUDIES ON AMERINDIAN

149

DOGS, 2

(Clutton-Brock et aE., 1976) may help to stimulate further biometric work, more especially on dogs. Bearing in mind these limitations, it is nevertheless possible to make some comment on the metrical variation seen in the Peruvian dogs, comparing them with other New World groups. Consideration of dental variation is left for a subsequent study. Considering the range of variation seen in the measurements (Table 4), the dogs may be considered of medium size. Head size was certainly much smaller than in wolf and larger modern dog breeds as exemplified by the Irish Wolfhound, St Bernard and Great Dane. Table 4. Ranges, means, standard deviations and coefficients of variation in the Peruvian series A measurements employed in the size and shape study. Also, comparable data for maximum cranial length in three series of early British dogs (Harcourt, 1974)

Maximum cranial length (1) Palate length (13) Bicondylar breadth (25) Bizygomatic breadth (30) Biorbital breadth (33) Palate breadth (34) Early

British

Range

No.

Z

0

cv

139-178 66-86 28-36 67-108 25-36 51-69

23 25 30 24 20 21

159.8 18.7 32.5 92.7 30.3 57.6

10.1 4.9 2.2 9.6 3.0 4.6

6.3 6.2 6.8 10.4 9.9 8.0

145-206 116-237 172-232

26 93 23

181 168 202

16.6 23.5 17.6

9.5 25.2 8.7

Dogs

Cranial length (1) Iron Age Cranial length (1) Roman Cranial length (1) Saxon

At the other end of the range, Peruvian and other Amerindian dogs were not as small as modern Pekingese and Griffon dogs. In scanning the metrical data as a whole, we have found it difficult to evaluate it in terms of Allen’s (1920) multiple categories“Plains-Indian”, “Larger Indian”, “Small Indian”, “Short-nosed Indian”, “Inca dog” and so forth. The one exception is the form originally named Canis ingae molossoides by Nehring (1887) and referred to by Allen (1920) as “Peruvian pug-nosed dog”. In the sample we have studied this group stands out even on visual inspection, especially by some noticeably reduced antero-posterior dimensions of the skull and in the more “S-shaped sagittal contour extending from the frontals on to the shortened nasals (evident in Figures l-4). Rather than use previous names we will refer to this form initially as group N. In considering series of domestic dogs, one is of course placed in the difficult position of evaluating a sample which may have been varyingly under the control of man, may have been fairly homogeneous-or perhaps was an assemblage of dogs used (and perhaps even selected) for more than one purpose. A further possibility is that the sample might display normal regional micro-evolution, but be influenced by superimposed relaxed selection-the product of the human protection of genetic variation which might not have survived in a non-domestic situation. With these points and problems in mind, the Peruvian craniometric data may be further considered. In Figure 6, two of the larger dimensions of the skull are plotted together. In the case of basal length and bizygomatic breadth, it was possible to view individual Peruvian and other Amerindian data against variation seen in modern dog varieties (provided by Stockard & Johnson, 1941), and also’ draw in a small sample of female wolves. It will be seen that the Peruvian and other Amerindian positions fall within a restricted area, with perhaps sufficient spread to suggest more than one homogeneous grouping.

150

D. BROTHWELL

ET AL.

l

I 70

5060

I 80

I 90

I I30

I I IO

I 120

I 130

B1zyg.Bd30)

Figure 6. Basal length (3) in relation to bizygomatic breadth (30) in Peruvian dog groups (A and N), modern wolf females, modern dog breeds and other Amerindian samples. The breeds are: (1) Saluki; (2) German Shepherd; (3) Pointer; (4) Great Dane; (5) St Bernard: (6) Bassethound; (7) Dachsund; (8) Cocker Spaniel; (9) English Bulldog; (10) French Bulldog; (11) Boston Terrier; (12) Pekingese; (13) Maltese Poodle; (14) Griffon. Symbols: X. means Wolf q’s;@-@ Dog breeds; 0, Peruvian group N ; :.I, Peru group A; .I., Other Amerindian dogs; A, lndian Knoll dogs.

l

I IO

I I20

I 130

I MO

I 150

I I60

I I70

I 180

Skull L (I)

Figure 7. Palate length (13) in relation to skull length (1) in early Peruvian dogs only. Ancon dogs are marked with an A in this scatter diagram.

STUDIES ON AMERINDIAN

DOGS, 2

151

It is interesting to note that the Amerindian dogs do not cluster about any one particular variety as defined today. Figure 7 considers total skull length against palate length for Peruvian specimens only. These two dimensions seem to be particularly useful discriminants for dogs, and it will be seen that there is a considerable difference in absolute measurements between A885 and A309 (Smithsonian specimens, collected by Hrdlicka). It is suggested that there are two groups represented here, namely a longer nosed form (group A) and the shorter nosed group N, with perhaps the dogs represented by 243, 161 and 162 being in an intermediate position. Palate lengths of the Peruvian sample were also considered with palate breadth against the perspective of some other Amerindian dog variation, as well as a small series of wolves (Figure 8). The wolf is well separated by palate length and breadth, as expected, but there is noticeable overlap with the range of variation in the non-Peruvian dogs.

1

I 60

I 70

I 80

I 90

I 100

I 110

I 120

I 130

I 140

Palate length (13)

Figure 8. Scatter diagram of palate breadth (34) against palate length (13) in three early Amerindian dog samples and in the wolf. Symbols: 0, Peruvian dogs; A, early Mexican dogs; 0, early dogs in U.S.A.; @, wolf individuals (male and female).

In the case of the skull index and palatal index (Figure 9), the additional dimension of palate breadth is included. It will be seen that two Ancon specimens diverge noticeably from the other Peruvian specimens for which the indices could be calculated. Considering the pattern of growth in the dog muzzle, there is clearly a high degree of correlation between longitudinal measurements of the palate and nasal bones. In plots of nasal length against both least orbital breadth and frontal breadth, the positioning of Peruvian skulls 635, 165 and 170 was sufficiently marginal to the others to suggest that group N measurements were beyond the variation expected to occur in an archaeological sample derived from a homogeneous series.

152

D. BROTHWELL ET AL.

I 60

I 70

I 80

I 90

I 100

I I IO

I 120

I 130

Palate index

Figure 9. Skull index in relation to palatal index for some modern dog breeds, two Peruvian groups and a general recent pariah dog series. Symbols :@-@. modern dog breeds; 0, Peruvian short-nosed dogs; A, other Peruvian dogs; Q, pariah dogs.

At present, comparative data for the lower jaw are very restricted. Figure 10 shows the relationship between ramus height and bizygomatic breadth (both influenced by masticatory demands). The wolf sample is extensive, whereas the Peruvian series is small. Nevertheless, a marked difference in size is again seen between dog and wolf, but unlike the other bivariate plots considered, there is no evidence of a separate N group. Because of the difficulty of finding comparative series of measurements for other groups of dogs, the analysis of a large number of measurements together seems for the present to be generally impossible. However, the literature (and photographic data) was scanned for measurements, and out of twenty skull measurements which could be equated with the von den Driesch list, six occurred commonly enough to be worthy of inclusion in a simple form of multivariate analysis. This permitted comparison with dog samples from Indian Knoll, Kentucky (Skaggs, 1946); a pooled early Mexican group (D.R.B., unpublished); a Woodland-Mississippi series (Haag, 1954); a general Kentucky sample (Haag, 1954), a general Peruvian sample (Smithsonian material), a Delaware series (Townsend site; Stewart, 1963); a New Mexico “Basket Maker” sample (pooled by D.R.B. from various sources); as well as samples of coyote, and wolf. A small general sample of pariah dogs was included, as an example of skulls of modern feral dogs with no advanced breed identity. A Great Dane and Fox-terrier were added as known variation in size. Finally, the new Peruvian craniometric data presented here and additional information given in the literature [especially Allen (1920), Nehring

STUDIES ON AMERINDIAN

DOGS, 2

0. 0 L

I 90

I loo

I I IO Blzygomotic

I

I

120

130

I

140

I

150

breadth (30)

Figure 10. Scatter diagram of ramus height (18) against bizygomatic breadth (30) for early Peruvian and wolf specimens. Symbols: 0, Peruvian dogs; $+ 9, individual male and female wolf specimens; 8, female wolf means (regional); 0, male wolf means (regional samples).

(1887) and Haag (1954)] were included as two groups; the short faced N group and a general medium size range of dog (group A). As the standard deviations for the six group A measurements are not high, they substantiate the view that this could be a single homogeneous variety of early Peruvian dog. Comparison with the standard deviations obtained by Harcourt (1974) for remains of British dogs (Table 4) supports this view. For the purpose of comparing the six dimensions together, the “size and shape” statistic of Penrose (1947, 1954) was used. Standard deviation units were calculated using group A standard deviations (in the absence of others). Although a consideration of the differences between all the groups would have involved the permutation of a large matrix, comparisons have been restricted to the two contrasting Peruvian groups (N and A) in relation to the others. Table 5 gives the basic information used, and Figures 11 and 12 the relationships of “size” and “shape”. It should be noted that unlike the usual procedure, positive and negative signs (established when d-values are summated) are retained. This assists in separating large and small groups when one or other distance value may seem similar. In Figure 11, marked size variation is clearly seen, and well separates wolf, coyote, Great Dane and Pariah series (large) and Indian Knoll females (small). Shape differences especially separate wolf males and females, as well as Peru series N. The Smithsonian

“Size” “Shape”

1 13 25 30 33 34

0.54 2.25

0.38 0.88

N-B

N-A

0.54 0.28

4.34 7.52

N-D 2.15 2.15

N-E

of group

0.17 0.10

A-F

1.32 2.56

N-F

Nfrom

0.38 0.02

A-G

A from

167.6 81.1 33.8 94.0 35.8 60.3

0.01 2.11

N-G

153.9 74.8 31.3 87.9 29.1 53.9

0.23 0.47

A-I

1.78 2.18

N-H

the rest

0.38 0.02

A-H

the rest

165.5 81.2 33.2 97.5 32.6 56.6

1.48 3.82

N-l

58.01 5.27

A-J

165.1 81.8 33.6 98.9 33.1 59.6

69.72 14.04

N-J

40.75 4.77

A-K

170.2 83.8 33.9 94.0 31.9 55.8

50.65 12.98

N-K

0.20 0.22

A-L

255.8 134.5 50.9 140.6 46.4 86.7

1.40 3.25

N-L

0.02 0.08

A-M

240.5 128.1 47.8 131.2 42.7 81.8

0.36 1.60

N-M

0.54 2.24

A-N

169.7 82.3 33.6 88.2 31.8 59.7

0.15 2.60

N-O

0.90 0.52

A-0

155.4 77.1 32.0 92.1 30.4 59.2

35.80 6.46

N-P

27.56 1.61

A-P

137.5 65.8 28.9 97.6 32.6 61.6

5.76 4.95

N-Q

2.78 0.64

A-Q

155.7 76.7 27.3 80.9 28.7 54.1

and the matrix

226.4 110.7 44.0 125.0 48.3 75.5

oj’size

184.5 90.3 37.6 101.2 35.5 59.5

and

*For brevity, these samples were assigned the letters A-Q, as follows: A=Pooled Peruvian group (excluding the short-nosed forms); B + C = Indian Knoll (6 and ‘2 groups), Skaggs (1946)‘s; D = Coyote $ + 2; E = early Mexican group (D.R.B., unpublished); F= Woodland-Mississippi dogs, Haag (1954); G =General Kentucky prehistoric, Haag (1954); H = Peru. Smithsonian sample, Haag (I 954); I = Townsend site dogs, Stewart (1963), J+K= Wolf. Southern USA. 6 and 9; L=New Mexico Pooled “Basket Maker” dogs (D.R.B.); M=Ancon dogs (this study+ literature); N = Peruvian short-nosed dogs; 0 = Fox terrier, modern; P= Great Dane: modern ; Q = Pooled sample, pariah dogs, [BM(NH)].

0.32 1.33

N-C

Distances

1.82 1.91

1.69 0.89

0.23 0.45

191.3 95.3 33.8 97.3 33.7 55.5

of group

133.8 69.9 30.2 77.6 26.6 59.7

A-E

Distances

142.3 75.4 31.7 88.3 30.7 59.7

A-D

159.8 78.7 32.5 92.7 30.3 57.6

Group*

5. Craniometric means employed in the Penrose “size” and “shape” comparisons, distances obtained jbr the Peruvian series A and N in relation to the rest

-~____-.-~--ABCDEFGHIJKLMNOPQ

A-C

A-B

Measurement number

“Size” “Shape”

~.__

-

Table shape

STUDIES ON AMERINDIAN

DOGS, 2

155

2AC

I-

@ ON

Group A\

“Shape” I

I

-6

I

-5

I

-4

I

-3

-2

-I

I

;GM.B

Q FA *A A Lg H EA

I I

I

I

2

3

Figure 11. “Size” and “shape” distances of various dog and wolf samples from the early Peruvian group A. Sample letters are defined in Table 4.

‘I Shape” I -8

I -7

-6

CA

GOUP

I

I

-5

I

I

-4

-3

I

0

I

G&2 0

I

-I f

I AB

A -I!--

,I

AL

AF @ A

H

-2E

-3-

+”

-4-

-5-

Jzl:’ J Figure 12. “Size” and “shape” distances of various dog and wolf samples from the early Peruvian group N. Sample letters are defined in Table 4. Symbols: A, non-Peruvian Amerindian dogs; A, early Indian Knoll dogs ‘(Kentucky); 0, Peru group A; a other Peruvian samples;@@ modern dog breeds: X, wolf 6 and 3; +, coyote.

156

D. BROTHWELL

ET AL.

dogs from Peru and the restricted Ancon sample were expectedly close to group A position. Other Amerindian dog series were relatively close together. Figure 12 shows the differences between the Peruvian group N and the other series. Marked differences in size and shape again separate wolf, coyote and pariah groups from the others. Although there is not such a marked size difference in the other groups, they generally diverge more from the group N position than in Figure 11. Shape variation is most pronounced, as regards Amerindian dogs, in the case of the Townsend Site (Delaware) and New Mexico series. Race, breed or morphotype As yet, we have been discussing size and shape in the skull without using these variations to classify the kinds of dog represented. It is of course reasonable to argue that although we can generally appreciate differences between wild canids and the variable domestic ones, the linking of samples of earlier dogs with modern breeds is unjustified and a waste of time (Burleigh et al., 1977). Obviously, modern breed differences are concerned with many non-skeletal characteristics, and in any case intensive selective breeding has accentuated even skeletal differences over the past few centuries. It is thus understandable that Harcourt (1974) in his review of early British dogs refrained from discussing the variation in relatian to similarities with breeds. Nevertheless, this does not exonerate us from the task of attempting to evaluate possible changes in dog populations through time, or looking for morphological differences in contemporary samples separated spatially. The calculation of means and standard deviations is in a way a classificatory act, giving information about an average “type” as well as the degree of heterogeneity in the sample. The application of multivariate “distance” statistics to osteometric data provides further systematic information without having to give false associations based on breed. There is, however, a need to bring some order to the infinite number of distances likely to result from statistical work, and in the case of Amerindian dogs each sample could display minor differences but still fall within a few separable groups. These groups we suggest, might usefully be evaluated as morphotypes. In the case of the Peruvian dogs, we have considered the degree of variation in the different measurements taken and feel that it is reasonable to conclude that two nuclear morphotypes are represented whatever the hybrid status may be of some of the dogs. Comparing the Peruvian dogs with some of the information provided by Stockard on “pure breeds” and using these purely as other morphotypes, we can begin to see where the Peruvian dogs are placed in the total range of dog variation. Regrettably only a few of Stockard’s measurements are comparable with those collected on the Peruvian dogs, and there is a great need for more dog measurements of comparative value. However, it can be seen in Figure 6, comparing bizygomatic breadth with skull base length, that wide variation exists in the triangle formed by the Griffon, St Bernard and English Bulldog. In this scheme, the Peruvian dogs span a noticeable part of the central area, perhaps suggesting well defined local micro-evolution (the Ancon and Mala groups for instance are well separated geographically). Unfortunately, what we consider to be a second Peruvian morphotype, which has previously been called a bulldog type, is only represented by a few specimens. This second morphotype is characterized particularly by the relatively short snout and the more “S”-shaped sagittal contour of the skull in the nasal and frontal area (Figure 3a and b). On the evidence of pottery and both early and recent literature (see Weiss, 1976), it is clear that we can recognize at least one more morphotype, that of the hairless dog, the result of canine ectodermic hypoplasia. Although this is the result of a dominant

STUDIES ON AMERINDIAN

DOGS, 2

157

mutant and therefore is not simply a normal variety of dog, at an osteological level it can be considered simply as another morphotype worthy of separate identification. On the other hand length of hair, well exemplified in Figures 1 and 2 of Ancon dogs, is difficult to evaluate in terms of the differentiation of separate dog stocks. Radiocarbon dating A total of eleven samples of keratin (hair and skin) taken from the mummified remains of domestic dogs from three archaeological sites in Peru has been dated at the British Museum radiocarbon laboratory (Table 6; see also Burleigh & Brothwell, 1978). The first seven dates listed in the table (BM-12361240, 1359, 1360) were obtained from material available in London. Most of this came from the Forbes Collection, formed c. 1913, in the British Museum (Natural History) and came originally from Ancon (11”45’S, 77”08’W). The remaining four dates (BM-1361-1364) were obtained from samples taken from specimens in the collections of the Laboratory of Palaeoethnozoology, University of San Marcos, Lima. These came originally from Chancay (11”36’S, 77”14’W) and Mala (12”4O’S, 76”36’W). Table 6. Radiocarbon sites in Peru

BM- No.

Site

dates for

Sample ref.

Weight (g) 6.3 4.0 (estimated) 13.0 6.0 5.5

1236 1237

Ancon Ancon

243 250

1238 1239 1240

Ancon Ancon Ancon

1359 1360 1361 1362

Ancon Ancon Mala Chancay

1363

Chancay

1364

Mala

251 635 729 (juvenile) DBa DBb Al57 Al62 (skull) Al62 (post-cranial) Al66

13.0 7.7 13.0 3.0

remains

of domestic

Radiocarbon age (relative to AD 1950) 757 f 48 bp (ad 1193) 834+ 88 bp (ad 1116)

dogs from

archaeological

Calibrated date (Clark, 1975) AD 125Ok 70 AD 1186?100

710? 41 bp (ad 1240) 1278 k 70 bp (ad 672) 2801+ 87 bp (851 bc)

AD 1303+ 65 AD 7OOk 85 1030+ 105 BC

949+ 687+ 1365k 1077 +

AD 1030+ 70 AD 1324+ 85 AD 626; 90 AD 912+130

50 67 77 122

bp bp bp bp

(ad (ad (ad (ad

1001) 1263) 585) 873)

2.0

142Ok221 bp (ad 530)

2.0

8392181 bp (ad 1111)

AD

571?230

AD 1180+ 190

As stated above the samples used for dating consisted of hair and skin but most were predominantly hair. In order not to damage the original valuable museum specimens seriously, only relatively small samples were taken. These ranged in weight from 2.0-13.0 g as shown in Table 6 and yielded amounts of benzene for measurement by the liquid scintillation counting method (Burleigh et al., 1976; Hall & Hewson, 1977) equivalent to between 0.42-3.40 g of carbon respectively. All the raw samples were well preserved and free from contaminants so that no chemical pre-treatment was required. The results are given in Table 6 in radiocarbon years based on the 5570 year half-life for l*C, the standard mode of reporting radiocarbon dates at present. The l*C measurements were corrected for isotopic fractionation so that the errors quoted with the dates are based on counting statistics alone and are equivalent to + 1 standard deviation (lo). None of the dates falls in a period in which the differences between radiocarbon and calendar years are large. For greater absolute accuracy, however, the equivalent calendar

158

D. BROTHWELL ET AL.

dates, allowing both for natural l*C variations and the more accurate half-life for lJC of 5730 years, are also given in Table 6. These figures were taken from the tables published by Clark (1975) but closely similar results would be obtained by using any of the other authoritative radiocarbon calibration curves and tables that have been published. The overall effect of these two corrections on this particular group of dates is small and the general pattern of the raw dates is not changed. The probable errors of the calibrated dates have been increased slightly over those of the corresponding radiocarbon dates to allow for uncertainties in the calibration curve itself. A number of comments can be made on the results. Firstly, with the exception of the single early date of 1030+ 105 BC (BM-1240), the dates do not fall into separate groups but appear to span a continuous period from about the 6th century to the early 14th century AD. This accords well with the archaeological evidence from all three sites. Secondly, the large gap of some 1600 years between the earliest date in the series (BM-1240) and the main sequence may perhaps indicate two separate periods of use of the necropolis at Ancon. Whether this is so or not the tradition of dog burial would appear on the evidence of these dates to have been a continuous one over a period of at least 2000 years. Finally, it should be noted that BM-1362 and 1363, from Chancay, are dates for different samples taken from the same animal. These two dates appear divergent but their errors are such that the difference between them, although considerable, is not significant statistically. Thus a weighted mean figure of about AD 807 _+1 18 should be taken for these two results. The history, and in particular the later history, of the domestic dog in the New World is already fairly well known (Clutton-Brock, 1977; Olsen, 1974; Olsen & Olsen, 1977; see also a forthcoming review by Dr Elizabeth Wing, Florida State Museum; Wing. 1977, personal communication). The earliest dated remains of dogs in North America are from Jaguar Cave, Birch Creek Valley, Idaho (Lawrence, 1967, 1968; Dort. 1975). charcoal from associated hearths having been dated by radiocarbon to 11,580 1250 bp (GX-395) and 10,370? 350 bp (r-400). Naturally it would be of the greatest interest to obtain direct dates for the bones of these animals (and this comment applies equally to the oldest domestic dog known from the Old World, from Palegawra Cave, northeastern Iraq, dated indirectly by radiocarbon to c. 12,000 bp) but at present this is an impossible ideal. Direct dating of such material will, however, eventually be possible by the method of radiocarbon measurement by high energy mass spectrometry now being developed, which will require only a few milligrams of carbon compared with the gram amounts needed for conventional radiocarbon dating. On the dating and other evidence available (see references cited in the preceding paragraph) fully domesticated dogs of Eurasian stock almost certainly came to North America with the people who migrated across Beringia during the later part of the Wisconsin glaciation, and must subsequently have spread fairly rapidly throughout the Americas. The earliest date for the presence of the dog in the central Andes of Peru is about 6000 bp on the basis of the radiocarbon chronology for the region (PiresFerreira et al., 1976), although canid remains which may represent domesticated animals have recently been found in a much earlier context (possibly c. 10,000 bp) at Lauricocha Cave. Again it would be of great interest and importance to be able to substantiate these early records by means of direct dates. From later periods there are numerous records for both North and South America, from a wide range of archaeological contexts. Many complete skeletons have been recovered and these show that great variation in the sizes and types of dogs existed. The present series of dates, while adding little that is new to the overall chronology, is of especial interest in that the large majority of the remains of domestic dogs known from archaeological sites in the New World and elsewhere are skeletal and remains

STUDIES

ON AMERINDIAN

DOGS.

2

159

which also include surviving skin and hair and soft tissues are very rare. Direct radiocarbon dates on well-preserved canid remains of this kind are therefore of particular value and there appears to be no other instance in which they have been obtained. Discussions and Conclusions

We have been able to study first hand, or collect together from other sources, information on 42 adult Peruvian dogs and some data on juveniles. Although attempts to consider this information in relation to other Amerindian dog series has been impeded by lack of comparable information, sufficient is published to suggest that local micro-evolutionary differences occurred and these of course may have been accentuated by the selection of varying qualities in the dogs by early Amerindians. In the case of Peru, as well as the abnormal condition of hairlessness, which seems to have attracted the interest of some earlier peoples there (but is not to be seen as a special “breed”) there appears to be a strong case for believing in long- and short-nosed varieties. On the limited evidence of preserved hair, short and long coats may have been another variable, possibly influenced by social selection. This dichotomy is best established so far in the material from Ancon (both by Nehring and the present study) and it can also be seen in the marked contrast in facial shape in skulls from Huacho, Peru, considered by Allen (1920, see his Plates 9 and 12). The noticeably undershot jaw seen in one of Nehring’s Ancon dogs, and in the BM(NH) Ancon dog 635, as well as in the Huacho specimen (USNM 176.307) is found today only in such breeds as the Boxer, Bullmastiff and Bulldog. But as Stockard & Johnson (1941) show, this undershot mandible may also occur in hybrids resulting from crosses between short-faced and long-faced forms. In view of this we feel justified in attempting to separate two series of dogs (A and N) in the Peruvian material-if only as an exercise in evaluating variation in dog samples. We realize, of course, that if we are correct in identifying two varieties, we may have placed in one or other category dogs which were in fact hybrids of both. In considering more than one form of Peruvian dog, we have been made aware of the temptation to use either previously assigned names for Amerindian dogs or to view them in relation to established modern breeds. There is no reason why we should not use both types of information as a perspective by which to view the Peruvian variation, but rather than employ such terms as “pug-nosed dog” we suggest that it is sufficient for the present, when considering archaeological samples, to attempt to fit the sample variation into a matrix of “morphotypes”. We therefore conclude that on available evidence, the Peruvian dogs show some genetic distinction in the form of morphotypes A and N, as outlined by us. Finally, we hesitate to suggest that this degree of differentiation is the result of intentional selection by prehistoric Peruvians, although it could have been. An alternative hypothesis might be that this degree of variation was already present in dogs moving with some early human groups into the Americas. It may be relevant to note that shortnosed dogs were common in China by 700 BC (Fiennes & Fiennes, 1968). Moreover, it has been suggested that an earlier form of what is now known as the Asian Molossus dog, could have been a forerunner of at least the modern short-nosed (and relatively broadfaced) varieties (Rine, 1965), if not perhaps a “mainstem” linked with the differentiation of other short-nosed morphotypes appearing in the archaeological record. Be that as it may, we can at least establish noticeable variation in early Peruvian dogs which seems to go beyond that expected to occur in one single evolving homogeneous population. Acknowledgements

We extend our appreciation to Juliet Clutton-Brock for her constructive criticism of this paper. Special thanks also go to Dr L. G. Lumbreras, Director of the Museo National

160

D. BROTHWELL

ET AL.

de Anthropologia y Arqueologia and to Dr J. L. Lorenzo, Departamento de Prehistoria, I.N.A.H., Mexico, who permitted accessto their zooarchaeological collections. Professor Gustav0 Or& of the Escuela Politecnica National in Quito, Ecuador, also very kindly permitted us to examine Peruvian dog material in his care. We are extremely grateful to Peter Dorrell, Noel Syers and the Photographic Studio of the British Museum (Natural History) for assistance in producing photographs and X-rays. Moira Mackenzie kindly produced the lateral skull drawings. One of us (D.B.) wishes to offer his special thanks to Heather Lownie, who patiently acted as interpreter on numerous occasions while in Peru and Ecuador. References Allen, G. M. (1920). Dogs of the American aborigines. Bulletin uj’the Museum of Comparative Zoology 63, 43 1-5 17. Allen, G. M. (1954). Canid remains from Pueblo Bonito and Pueblo de1 Arroyo. Smithsonian

Miscellaneous

Collections

12.4, 385-389.

Burleigh, R. & Brothwell, D. (1978). Studies on Amerindian dogs, 1: Carbon isotopes in relation to maize in the diet of domestic dogs from early Peru and Ecuador. Journal of Archaeological

Science 5, 355-362.

Burleigh, R., Hewson, A. & Meeks, N. (1976). British Museum natural radiocarbon measurements VIII. Radiocarbon 18, 16-42. Burleigh, R., Clutton-Brock, J., Felder, P. J. & Sieveking, G. de G. (1977). A further consideration of Neolithic dogs with special reference to a skeleton from Grime’s Graves (Norfolk) England. Journal of Archaeological Science 4, 353-366. Clark, R. M. (1975). A calibration curve for radiocarbon dates. Antiquity 49, 251-266. Glutton-Brock, J. (1977). Man-made dogs. Science 197, 1340-1342. Glutton-Brock, J., Corbet, G. B. & Hills, M. (1976). A review of the family Canidae, with a classification by numerical methods. Bulletin of the British Museum (Natural History) Zoology Series 29, I 17-199. Colton, H. S. (1970). Aboriginal Southwestern Indian dogs. American Antiquity 35, 153-159. Dort, W. (1975). Archaeo-geology of Jaguar Cave, Upper Birch Creek Valley, Idaho. Tebiwa 17, 33-57.

Driesch, A. von den (1976). A guide to the measurement of animal bones from archaeological sites. Peabody Museum Bulletin 1, 135 pp. Fiennes, R. & Fiennes, A. (1968). The Natural History of the Dog. London: Weidenfeld and Nicolson. Giles, E. (1960). Multivariate analysis of Pleistocene and recent coyotes (Canis latrans) from California. University of California Publications in Geological Sciences 36, 369-390. Haag, W. G. (1948). An osteometric analysis of some aboriginal dogs. University of Kentucky Reports in Anthropology 7, 264 pp. Haag, W. G. (1954). A mummified dog from the Lighthouse Site, Supe. In (G. R. Willey & J. M. Corbett, Ed.) Early Ancon and Early Supe Culture. New York: Columbia University Press, pp. 138-140. Haag, W. G. & Heizer, R. F. (1953). A dog burial from the Sacramento Valley. American Antiquity

18, 263-265.

Hall, J. A. & Hewson, A. D. (1977). On-line computing and radiocarbon dating at the British Museum. Journal of Archaeological Science 4, 89-94. Harcourt, R. A. (1974). The dog in prehistoric and early historic Britain. Journal of Archaeological Science 1, 151-175. Jolicoeur, P. (1959). Multivariate geographical variation in the Wolf Canis lupus L. Evolution

13, 283-299.

Kidder, A. B. & Guernsey, S. J. (1921). Papers of the Peabody Museum of American Archaeology and Ethnology. Harvard 8,44 pp. Kurten, B. (1955). Sex dimorphism and size trends in the Cave Bear Ursus spelaeus. Rosenmtiller and Heinroth. Acta Zoologica Fennica 90, 148.

STUDIES ON AMERINDTAN

DOGS, 2

161

Lawrence, B. (1967). Early domestic dogs. Zeitschrtft fir Saugetierkunde 32, 4459. Lawrence, B. (1968). Antiquity of large dogs in North America. Tebiwa 11, 43-49. Lawrence, B. & Bossert, W. H. (1967). Multiple character analysis of Canis lupus, lutruns and familiaris, with a discussion of the relationships of Canis niger. American Zoology I, 223-232. Nehring, A. (1884). Ueber Rassebildung bei den Inca-hunden aus den grabern von Ancon. Kosmos 15, 94-l 11. Nehring, A. (1887). Mammals. Section 15. In (W. Reiss & A. Sttibel, Ed.) The Necropolis of Ancon in Peru Vol 3. Berlin: Asher. Olsen, S. J. (1974). Early domestic dogs in North America and their origins. Journal of Field Archaeology

1, 343-345.

Olsen, S. J. (1976). The dogs of Awatovi. American Antiquity 41. Olsen, S. J. & Olsen, J. W. (1977). The Chinese wolf, ancestor of New World dogs. Science 197, 533-535.

Penrose, L. S. (1947). Some notes on discrimination. Annals of Eugenics, London 13, 228-237. Penrose, L. S. (1954). Distance, size and shape. Annals of Eugenics, London 18, 337-343. Pires-Ferreira, J. W., Pires-Ferreira, E. & Kaulicke, P. (1976). Preceramic animal utilization in the central Peruvian Andes. Science 194, 483490. Rine, J. Z. (1965). The World of Dogs. New York: Doubleday. Ritchie, W. A. (1945). An early site in Cayuga County, New York. Type component of the Frontenac Focus, Archaic pattern. Researches and Transactions of the New York State Archaeological Association 10, 158 pp. Skaggs, 0. (1946). A study of the dog skeletons from Indian Knoll with special reference to the coyote as progenitor. Indian Knoll, Site Oh 2, Ohio County, Kentucky (Webb, W. S.). University of Kentucky Reports in Anthropology 4. Stewart, T. D. (1963). Skeletal remains of aboriginal dogs. The Archeolog 15, 54-58. Stockard, C. R. & Johnson, A. L. (1941). Section III. In (C. R. Stockard et al., Ed.) The Genetic and Endocrinic Basis for Differences in Form and Behavior. Philadelphia: Wistar Institute. Stockdale, A. (1917). The hairless dog. Journal of Heredity 8, 519-520. The, T. L. & Trouth, C. 0. (1976). Sexual dimorphism in the basilar part of the occipital bone of the dog (Canis familiaris). Acta Anatomica 95, 565-571. Weiss, P. (1976). El perro Peruano sin pelo (perro Chino, viringo, Ccala and Ccalato). Publicaciones

de1 Muses National

de Antropologia

y Arqueologia.

Serie: Paleobiologiu

1,

33-54. White, T. E. (1955). The dog bones from the Buffalo Pasture Site. American Antiquity 21, 170-178. Wing, E. S. (1975). La domestication de animales en 10sAndes. Allpanchis 8, 25-44. Young, S. P. & Goldman, E. A. (1964). The Wolves of North America. New York: Dover.