Stanley J Olsen. Cambridge World History of Food. Editor: Kenneth F Kiple & Kriemhild Conee Ornelas. Volume 1. Cambridge, UK: Cambridge University Press, 2000.

The Dog as Human Food

The difficulty that confronts one at the very beginning of a study of the dog as human food is the lack of convincing evidence relating to the use of dogs (Canis familiaris) as food by early humans. Most of the discussions in published articles are either speculative or employ relatively recent evidence of dog consumption and extrapolate these data to the distant past.

There are some fine published discussions that relate to the eating of dogs in more recent times. One of the best is by Margaret Titcomb (1969), who documents the importance of the dog to human cultures in the ancient Pacific area. Titcomb believes (as do I) that the dog of the Pacific Islands was originally carried by canoe from the Asian mainland along with the pig and chicken for food on the voyage.

It was fed carefully upon vegetables such as taro and breadfruit so that its flesh was palatable to its human owners. Captain James Cook noted in his journal of 1784 that dogs were fed and herded with the hogs. Titcomb (1969) relates that they were fattened on poi and that an informant remembered that relatives penned up puppies to keep them from eating unclean food. They were fed three times a day on poi and broth until fattened. Natives considered baked dog a great delicacy, but they never offered it to foreigners, who held dog meat in great abhorrence, according to the wife of a missionary in 1820.

Preparation of dogs as food is abstracted by Tit-comb (1969), who writes: “Like the pig, the dog was killed by strangulation in order not to lose the blood. The blood was poured into a calabash (gourd) used for cooking, and red hot stones were added, causing the blood to coagulate; it was then ready to eat.” The meat might be baked or cut up and wrapped in ti leaf bundles and cooked in boiling water. An informant stated that the upper part of the backbone was preferred by many diners, although some liked the ribs best, and others preferred the brain. The latter has been attested to by dog skulls, with holes in the crania, found in archaeological excavations.

There is some indication of the numbers of dogs that were used as food from the dance anklets, made of dog teeth, that survive in museum collections. Dog canine teeth were drilled for attachment to fabric bands and worn as dance rattles. One, in the Bishop Museum, has 11,218 teeth; it has been estimated that 2,805 dogs were required to produce the artifact.

Domestic dogs found their way from island to island across the Pacific Ocean. A major reason for their success as immigrants in the region was a total absence of competition from wolves, jackals, and foxes. Pacific Island canids themselves did not become feral, probably due mainly to a lack of adequate food in the wild, which meant they had to rely on their human owners for subsistence.

Today dogs are employed as food for human consumption in many parts of Asia, and China in particular, but the origins and reasons for this practice are not well documented, or if they are, they are probably recorded in one or more of the many Asian sources not yet translated.

Dog Fossils and Their Interpretation

Dog consumption practices in the Pacific and Asia notwithstanding, the dog’s main attraction to humans, historically, has been its ability to hunt other animals selected as food by humans rather than serving as a source of food itself.

Over the past 50 years that I have been examining animal bones from archaeological sites, I have yet to see the bones of dogs bearing cut or butcher marks that would indicate that dogs had unquestionably been butchered for food. In the southwestern United States, for example, and particularly in Arizona, there are literally hundreds of articulated prehistoric dog burials and many more isolated postcranial bones. The latter often show evidence of cutting, but it is evident that they were being prepared for bone artifacts. Most represent the shafts of humeri or femora and were scored by flint blades in transverse directions below the proximal articular condyle or the distal articular condyle, resulting in a bone tube when the articular ends were removed. Many of these artifacts, or “tubes,” have been recovered from archaeological sites along with the discarded ends of the bones.

Some archaeologists have used the evidence of burned or charred dog bones to indicate meal preparation. However, experiments with fresh material have proven that burned or charred bone is the result of grilling to the point where edible flesh has been completely destroyed. Indeed, it must be destroyed before any change is apparent on the bones.

Generally, the sample size of dog burials is inadequate to determine their value to the associated human population. However, there are several occurrences of multiple dog burials that suggest reasons other than culinary for the buried animals. In Askelon, Persia, 1,238 dog skeletons of all ages, associated with a period dated between 822 and 428 B.C., were excavated and examined by archaeologists (Wapnish and Hesse 1993). There was no evidence of butchering or food preparation associated with these dogs.

Similarly, a site of puppy burials from the first-century-A.D. Roman villa at Lugnano, Italy, was excavated in June 1992. At least some of these dogs were associated with child burials, which may have had some ritual significance. In fact, dogs in association with human burials are not uncommon. Intentional burials of dogs, both separately and with humans, are known from Chinese sites. Some of the dogs are in the same coffins with the humans; others have their own separate burial boxes.

Indeed, there is always a danger of misinterpreting dog remains when only bones are present. One of the best examples of this potential for misinterpretation is connected with a 1901 photograph taken on the shore of Penzhina Bay, in eastern Siberia on the Sea of Okhotsk. The photograph shows four large sled dogs hanging by their necks atop long poles that were set into the frozen earth (Perkins 1981). Waldemar Jochelson, the photographer (on what was an American Museum of Natural History expedition), explained:

All along the bank of the Paren River were stakes driven into the snow, with dogs hanging on them, their muzzles pointing upward. In the light of the spring sun, this long row of dog-sacrifices offered a queer and sad sight. I found out that the greater part of the dogs had been killed by the drivers of the sledges in gratitude for their safe return from Gizhiginsk, and to guard their villages from the measles, which a year previous had come to them from that little town. (Perkins 1981)

The dogs were eventually cut down from the poles and buried, and there was no evidence left indicating the circumstances that led to their deaths. It is doubtful that any excavators of the dogs would be able to glean the method of their execution or the reason for their demise. The point is that ritual sacrifices of dogs are not generally recorded. One example is a fenced brick-lined pit, off to one side of the market in Kathmandu in Nepal, whose purpose is for the sacrifice and dismembering of dogs. If it were not for the tourist photographs taken when the rituals are in progress, the small fenced pit might be mistaken for a dry fountain.

At sites in Egypt and prehistoric southwestern North America, dog mummies have been recovered. In the former region, humans did the mummifying, whereas natural causes were responsible in the latter. From these discoveries we can derive information about dog size and color of pelage and can make comparisons with surviving breeds of dogs. Two of the best-preserved animals are a small black-and-white dog and a yellowish-white one from a Basketmaker site at Marsh Pass, Arizona.

These dogs were mummified naturally in the hot, dry environment and placed in White Dog Cave with a mummified Pueblo Indian wrapped in a woven shroud. The dogs were a bit longer legged than the mongrels found in the vicinity today, but the color patterning and small size are rather common in the general area of Marsh Pass.

Dog Domestication

If the pages of the past are blurred with respect to the use of the dog as food, they are equally hard to read in matters of dog domestication, which constitutes the burden of most of the remainder of this chapter. Much of what has been and is being published on the events leading to canid taming and domestication is based more on supposition than on hard evidence. Woven into what follows is my own version of this process (based on Olsen 1985).

Studies relating to the social structures of both humans and wolves shed some light on what may have taken place in the initial development and relationship of these two groups of early hunters that eventually culminated in the close relationship between dogs and humans prevailing today throughout much of the world. Foremost among these studies are those by L. D. Mech (1970), M.W. Fox (1971), and B. H. López (1978). Perhaps the best compilation that relates to this relationship is a series of papers edited by R. L. Hall and H. S. Sharp (1978). Several theories dealing with the attitudes of humans toward wolves and of wolves toward humans are discussed in detail in their volume.

Human hunter–gatherer societies and wolf packs were similar in a number of respects. Both comprised social units that were relatively small in number. Both were capable of hunting over open ground or wooded areas, pursuing rather large game and exerting considerable physical effort and energy over prolonged periods to accomplish their goals. Both used hunting methods that required a pack or team effort as opposed to the tactics of a lone hunter.

Faunal evidence from paleoanthropological sites of the late Pleistocene age (c. 10,000 B.C.) indicates that Homo sapiens, at this stage of development, depended on a diet that included a sizable portion of game animals. As carnivores, the wolves of this period also depended on wild game, and because hunting game by running it down takes a good deal more time than consuming the kill, one of the characteristics of both these predator groups would have been sharing the spoils of the hunt. Such socialization would have developed independently among both wolves and humans (Hall and Sharp 1978).

Another social feature shared by wolves and humans would have been that gender did not necessarily determine who did the hunting. Rather, such a determination would have depended on the individual’s maturity and ability to hunt successfully. The nonhunters of the pack (immature animals), therefore, would gather in rendezvous areas that were selected for regrouping and for the sharing of freshly killed game provided by the active hunting members of the pack (Mech 1970; Hall and Sharp 1978; López 1978). The wolf packs’ methods of obtaining game appear to be well organized and planned and should be briefly abstracted here as a possible model of how human hunting societies may have obtained game during the Pleistocene.

As pack hunters, wolves are able to capture and kill animals much larger than themselves. Artiodactyls such as deer, mountain sheep, elk, caribou, and even the moose, a formidable adversary, are all chosen prey. Rabbits, beaver, and (at times) fish in shallow water, as well as other small mammals, are also caught and eaten (Mech 1970). Wolves have been observed to change their hunting tactics more to suit the terrain and weather conditions than to suit the habits of a particular animal that they are pursuing. Their ability to procure food is not always successful; rather, feast or famine may be more the norm for these animals.

The selected prey in an ungulate herd may be an old or weakened animal, but this is not necessarily the case. Healthy, prime adults, as well as immature animals, are also selected and killed by wolves (Mech 1970). The method of bringing down such an animal is for the wolf to run alongside, slashing and tearing at its hindquarters, flanks, and abdomen until it is weak enough from its injuries to be dispatched (López 1978). Wolves may grab an animal, such as the moose or caribou, by the nose, holding on, while other members of the pack close in for the kill (Mech 1970; Hall and Sharp 1978; López 1978).

It is the case, however, that wolves generally choose animals that are easier to procure than healthy, active prey. Animals that are injured, infected by parasites to such a degree that they are weakened, or foundering in deep snow are all known to have been selected by wolf packs for food (López 1978). Animals in this condition, of course, signal their plight to some degree; these signals would not be lost on keen observers, such as wolves (or humans), and the weakened prey would be relatively easy to dispatch.

If a prey animal runs, it is certain to be pursued, especially if it is a lone individual (Mech 1970). To weaken such an animal, particularly if it appears to the wolf to be a strong and healthy adult, the practice of relay hunting may be put into practice, taking advantage of the terrain to wind the pursued. The hunt may end in a rush, terminating the animal’s struggles in a few moments, or the dogged pursuit may go on for miles before the pack closes in. Alternatively, a wolf pack of four to six animals may send out two wolves to herd a lone victim into an ambush of the remaining members of the group (Mech 1970).

Wolves that have maintained a territory for a considerable length of time may know the terrain well enough to use shortcuts to intercept running prey. They are also known to hunt into the wind when approaching a feeding herd of artiodactyls.

The similar pack-hunting methods and social structures of early humans and wolves would not, in themselves, have been responsible for the first taming and domestication of the wolf. Perhaps humans observed the wolf’s ability to track successfully by scent, or its ability to cover ground swiftly enough to elude human pursuit. It is possible that hungry wolves were enticed (not necessarily by human design) to come close to a campfire, where meat was being cooked and the refuse discarded in the immediate vicinity of the camp. Perhaps wolves that had attached themselves loosely to human habitation areas would consider such camps as their home territory, and their warning growls toward intruders would also warn the human inhabitants of the approach of such outsiders. If such an association occurred, it is not unreasonable to assume that these events would bring about firmer ties between wolves and humans to their mutual advantage. Unfortunately, evidence of the kind needed to prove or disprove this speculation is not the sort that could be preserved in the archaeological record.

If a hunter were to have initiated a closer association with wolves by taking a wolf pup from its den (possibly if a pup’s parents were killed), this action would have encouraged imprinting of the substitute human family leader on the young wolf’s behavior pattern as it developed and matured in close association with the human hunting group. Thus, J. P. Scott (1950) wrote:

Behavior studies seem to indicate that the reason for the apparently easy and successful domestication of the wolf lies in the fact that it is a species with highly developed patterns of social behavior and that a large number of these patterns are sufficiently similar to human ones to permit mutual social adjustment between man and wolf. The primary stimuli are so similar in the two species that appropriate and recognizable social behavior is evoked.

But it is also quite possible that there existed a mutual respect between the wolf and human of 10,000 to 15,000 years ago for the hunting prowess of the other. At such a time, humans were themselves probably as rugged and aggressive as the wolves with which they shared a hunting territory.

In order for any animal, including wolves, to be successfully tamed and domesticated, it must be able to suppress its natural or normal living pattern and subjugate it to that which is dictated to it by humans. The wolf (in most instances), and the resulting domestic dog, have been able to do this, as have all of the common domestic animals that are familiar to us. But some animals cannot cross this bridge from wild to domestic. Records extending as far back as early dynastic Egypt indicate that many animals that seemed potential domesticates did not fulfill that potential. Some Egyptians experimented with hyenas, gazelles, and foxes, but they had little success in molding them into household animals. Similarly, the cape hunting dog (not a dog in the true sense of the word), the coyote, the jackal, and all of the large, wild felids were never brought beyond the tame or semi-tame stage. The modern wolf, however, appears to be an animal that will easily and fully accept the companionship of humans in a manner that nearly approaches that of a domesticated dog.

Since we are discussing the possibilities of the first friendly contacts between wolves and humans, it is appropriate to mention at this point the evidence we have for early domestication through osteological changes in the wolf. It is generally known and accepted by workers concerned with early canid domestication that the first observable changes take place in the skulls, jaws, and dentition of these tamed canids. More specifically, these alterations include the foreshortening of the muzzle, or rostrum, and the crowding of the tooth rows, along with a comparative overall reduction in the size of the teeth.

As domestication occurred, the mandibles deepened midway along the horizontal ramus, with a more convex inferior margin than that found in similar-sized wild wolves. These are all characteristics found in most domestic dogs of wolf size.The coronal apex takes a decided backward turn, a condition found in some Chinese wolves (Canis lupus chanco) but generally absent in wolves from other geographic areas (Olsen and Olsen 1977). The foreshortening of the muzzle, or rostral, area of the skull has been observed in the remains of wolves recovered from geographically widely separated paleontological and archaeological sites of late Pleistocene age and later.

One of the earliest documentations of this fore-shortening of wolf skulls was made by J. G. Pidoplichko (1969), who described short-faced, and possibly tamed, wolves. Their fossils were collected from excavations conducted in 1954 at a late Paleolithic campsite of mammoth hunters at Mezin in the Chernigov region of the Ukraine. They were found in association with a lithic assemblage amid the remains of 116 individuals of woolly mammoths (Mammuthus primigenius). The remains of wolves with proportions of those living in that area today (Canis lupus albus) were also found, along with the three or four short-faced animals.

Pidoplichko coined the taxonomic designation Canis lupus domesticus for these short-faced wolves. I, however, would propose to change the taxonomic assignment to Canis lupus familiaris for these and other similarly differing wolves and to use the name designated for the domestic dog to identify this Pleistocene race or subspecies, rather than create a new trinomial designation. The use of familiaris follows the correct taxonomic progression from a wild wolf, through taming, to the domestic dog. This term seems to be less confusing than the name domesticus, which has commonly been used as the scientific name for the domestic cat (Felis domesticus, now Felis cattus).

For a number of years, particularly during the 1930s, the Frick Laboratory of the American Museum of Natural History in New York City had field representatives stationed in the Fairbanks area of Alaska. They were there to cooperate with the University of Alaska and the Fairbanks Exploration Company in collecting faunal remains as they were uncovered in muck deposits by hydraulic gold-mining operations. Such placer-mining methods, however, made it impossible to determine the stratigraphic context of a particular bone or artifact, because many had been collected by the gold-dredge crews from the spoil dumps of reworked and discarded matrix.

In fact, there is a general consensus that artifacts from such areas, which may originally have been on the surface, were subsequently relocated several meters below the ground level and mixed with the bones of extinct animals (Rainey 1939). Thus, it might have been possible to find some artifacts in association with the older, extinct animals if different methods of mining had been practiced when the bulk of known Pleistocene vertebrate material was collected in the 1930s.

Between the years 1932 and 1953, 28 more or less complete wolf skulls were obtained from the muck deposits in an area north and west of Fairbanks. The age of the deposits from which these wolf skulls were obtained has been determined as Upper Pleistocene, Wisconsin Age (about 10,000 B.C.). The geologist T. L. Péwé published two reports (1975a, 1975b) on the geology and stratigraphy of this area and the associated recovered fauna. The wolf skulls were collected from dredging operations on Ester Creek, Cripple Creek, Engineer Creek, and Little Eldorado Creek. All of these creeks also yielded Paleolithic artifacts (Rainey 1939). However, the association of these artifacts with the skeletal remains of extinct animals is problematical, because the mechanical mixing of artifacts and bones, just discussed, prevented the collectors from obtaining any evidence that might have existed of a contemporary association of Pleistocene wolves with humans.

T. Galusha, of the Department of Vertebrate Paleontology at the American Museum of Natural History in New York City, worked for many years on the wolf remains from the Fairbanks dredging operations. He noted that a number of the skulls were extremely short-faced for wild wolves and approached modern Eskimo dogs in facial proportions, although they were considerably larger animals overall.

He turned the project over to me shortly before his death, and I continued his studies and compared this collection with a large series of both Pleistocene and Holocene wolves as well as with Eskimo dogs from both Greenland and Siberia. I feel confident at this point in stating that the short-faced wolves from the Fairbanks area appear to be forerunners of the later, domesticated Eskimo dogs.

The proportions of the skulls of these wolves that vary do so in the rostral area. The area of the skull that is anterior to the infraorbital foramen is noticeably foreshortened and constricted laterally in several of the skulls. Two of them (F:AM70933 and F:AM67156) lack anterior premolars. Several others have the full complement of teeth in the abbreviated dental margin. The dentition of F:AM67153 is considerably smaller than the average wolf of that area and approaches the overall tooth size found in the Eskimo dog.

Dishing of the rostrum, when viewed laterally, is evident in all of the short-faced skulls identified as Canis lupus from the Fairbanks gold fields. The occipital and supraoccipital crests are noticeably diminished compared to those found in average specimens of C. lupus. The occipital overhang of these crests, a wolf characteristic, is about equal in both groups of C. lupus. Multivariate analysis of all of the wolves from the Fairbanks collection separated the collection into two groups, based on either their predominant wolf-like or dog-like characteristics. Yet, as has already been noted, it may never be possible to establish if there was a close human association with the Alaskan Pleistocene wolves, because of the types of deposits in which both the wolves and the lithic assemblages occur and the conditions under which both were collected.

In 1939 and 1940 an Eskimo village site was uncovered that dated from the first millennium B.C. This village, named Ipiutak, is believed to represent an occupation of early Eskimo immigrants from Siberia. The artifacts suggest a Neolithic culture (Larsen and Rainey 1948). The site is located on the tip of Point Hope, Alaska, approximately 125 miles north of the Arctic Circle, and is at the westernmost point of the continent north of the Bering Strait.

At least five more or less complete Canis skulls were recovered from the archaeological excavations at Ipiutak. These were studied and reported on by O. J. Murie in an appendix to the site report by H. Larsen and E. Rainey (1948). After comparing these skulls to those of wolves, Eskimo dogs, and other large domestic dogs, Murie decided that they fell into the Siberian class of dogs. Any slight differences between the Ipiutak dog skulls and the Siberian and Alaskan dog skulls with which they were compared were within the size variation considered to be normal for domesticated dogs of this type.

One animal in particular (No. H43) was considered to be a dog–wolf hybrid, as the large size and heavier mandible and teeth approached those of a wolf rather than a dog. Murie concluded that two of the skulls represented the so-called Siberian husky type, three may have been variants of this same type, and the other animal (No. H43) was the hybrid just mentioned. He believed that there was adequate evidence that the Ipiutak dogs were of Asian origin.

Both sides of the Bering Strait have seen numerous explorations and excavations searching for the earliest evidence of habitation sites of the Asian migrants who reached the North American continent. Unfortunately, very little relating to vertebrate remains has been reported. This may be partly because of the poor preservational conditions that characterize this region and that have resulted in an extremely small sample upon which to base our interpretations. However, the subordinate role of organic remains in many early archaeological interpretations has also had a great deal to do with this gap in our knowledge of fossil vertebrates of all taxa from archaeological contexts.

An example of this subordinate role can be seen in the 331-page report on the archaeology of Cape Denbigh (Giddings 1964). The text and 73 plates are devoted to a thorough coverage of nearly all aspects of the excavations on Cape Denbigh, but the recovered faunal remains are listed in a table of only seven lines in one paragraph. No scientific taxonomy is given for these animals, and some remains are classified simply as “bird” or “other.” Obviously, it is quite possible to overlook many comparatively small canid fragments if one’s interests are funneled in other directions. It is also possible that canids were not collected because it may have been assumed that they represented the common, local wolves, when in actuality they might have represented primitive dogs, as well as wolves or crosses between wolves and dogs.

An intriguing collection of seven large canid skulls that may represent both domestic dogs and wolves or even wolf–dog crosses were found at the Bagnell site, which is a two-component village situated on the edge of a terrace above the Missouri River flood plain just north of Sanger, North Dakota. The dates of the proveniences from which the skulls were collected are from as early as circa A.D. 1590, with the latest occupation from the earliest layer yielding two skulls of domestic dogs of wolf size. Two more of the skulls are definitely those of wolves, but from the later time of occupation; one skull is of a domestic dog, also of wolf size, and the remaining two are of wolves. A most intriguing observation is that one of the domestic dog skulls from the early layer possesses two strong wolf characteristics and seven morphological characteristics that are typical of large, domestic dogs. Admittedly there is a problem of temporal control for these proveniences; however, at this time it does not seem likely that the earliest proveniences contain any intrusive mixtures of dogs that were introduced by Europeans. Large, wolf-sized dogs, particularly those that appear to be wolf–dog hybrids, are so rare in an archaeological context that these Bagnell canids are worthy of considerable note.

Another published account of a large, hybrid cross of a wolf and a dog is that contained in the report of excavations of a bison-kill site in southwestern New Mexico (Speth and Parry 1980). This find, a single canid skull, is illustrated and analyzed in the report, but unfortunately no provenience is given for this specimen. The overall range of dates for all proveniences on the site are from as early as A.D. 1420 ± 125 to A.D. 1845 ± 100.This canid would be of interest in relation to early taming of the wolf and, perhaps, its crossing with smaller Indian dogs, but only if it could be established that it was collected from a provenience that predated the European introduction of large mastiffs or hounds. Since the range of dates is so great, it could easily be the result of a more recent crossing of a European dog and local wolf. (The occurrence is recorded here as a plea to future workers to include all pertinent data and not just that which pertains to comparative zoology.)

To date, the oldest substantiated finds of prehistoric domestic dogs in North America are those from Jaguar Cave, Idaho. Fossil fragments, consisting of incomplete mandibles, a single left mandible, and a small portion of a left maxilla, were all collected from excavations in an early Holocene rock-shelter in the Beaverhead Mountains of Lemhi County, Idaho. The excavations were conducted by a joint expedition of the Peabody Museum of Archaeology and Ethnology at Harvard University and the Idaho State Museum at the Idaho State University in 1961 and 1962. Carbon-14 (14C) dates indicate that the age of the deposits ranges from about 9500 to 8400 B.C. (Lawrence 1967).The site was determined to be a hunting camp (Sadek-Kooros 1972).

An interesting aspect that remains unexplained concerning the canid from this site is its unwolflike appearance (Lawrence 1967), which suggests there may be an even earlier form, as yet undiscovered, that may link these Jaguar Cave dogs with a wild, ancestral form. One would not expect to find these early dogs in a locality so far south as the Jaguar Cave rock-shelter without finding similar remains in sites closer to the Bering Strait. The morphological features that are present in the Jaguar Cave dog fragments are characteristically the same as those that are present in known specimens of C. familiaris. For example, critical measurements taken of all fragments indicate that they are too small to be derived from the wolf. They were then compared to the coyote, but they differ from this smaller, wild canid in being more massive, deeper dorsoventrally, and thicker lateromedially. The tooth rows are short when compared to the size of the individual teeth, and this shortening of the jaws is accompanied by crowding of the tooth row. This aspect is particularly noticeable in the area of anterior premolars, where the alveoli do not lie in a straight line; they are, rather, set obliquely. The muzzle, as far as could be projected from the fragments, is more shortened than in the coyote, and this characteristic seems to be well developed, as in later domestic dogs.

Some lower jaws of what are surely coyotes also occurred in the same site. These were determined to be of the local subspecies Canis latrans lestes. Canid material from a considerably later site (also in Birch Creek Valley, Idaho), having a 14C date not earlier than 2500 B.C., was also described by B. Lawrence (1967) as belonging to domestic dogs. The specimens included four nearly complete mandibles, a number of mandibular fragments, one broken skull, and two cranial fragments. The specimens from this later site were determined to be very similar in form and size to Eskimo dogs, with which they were compared in the Museum of Comparative Zoology at Harvard University. By inference, the specimens from Jaguar Cave were also from this group of dogs.

There exists a considerable temporal gap between the Jaguar Cave dogs and later prehistoric dogs from other areas in North America. Fossils from areas in the Southwest date from at least the time of Christ (Guernsey and Kidder 1921). From other areas they date perhaps from 8400 B.C. (McMillan 1970). A dog from the Koster site in Illinois (Struever and Holton 1979) has an assigned date of 6500 B.C., and dogs from White’s Mound in Richmond County, Georgia, have been given a date of about 500 B.C. (Olsen 1970). F. C. Hill (1972) briefly noted a Middle Archaic dog burial from Illinois with an accompanying date of 5000 B.C.

G. M. Allen (1920) published the first comprehensive discussion of early dogs that were associated with the prehistoric peoples of the Western Hemisphere. This publication, unfortunately, is now out of date as well as out of print. Most of the critical finds relating to the development of the domestic dog were made subsequent to Allen’s work. W. G. Haag (1948) based his evaluations of an osteometric analysis of some eastern prehistoric dogs on Allen’s early publication, but Haag’s monograph is also out of date.

The monograph by Allen was for many years the standard work that was used to classify domestic dogs found in association with prehistoric human cultures. The author listed these dogs as falling into the following groups, determined by comparative skull measurements and form: “(1) a large, broad-muzzled Eskimo dog, (2) a larger and (3) a smaller Indian dog, from which are probably to be derived several distinct local breeds. Of the larger style of dog, as many as eleven varieties may be distinguished; of the smaller, five” (Allen 1920: 503).

Some years ago I placed southwestern prehistoric dogs into two assemblages, small and large, following, more or less, Allen’s classification. The animals in the small group were fox terrier–sized and were further refined or separated into small, short-faced, and small, long-faced dogs. The animals in the large group were long-faced and were comparable in size to the local coyote but a bit heavier in overall proportions. This latter group was referred to by Allen as the large Pueblo Indian dog, or the Plains Indian dog.

However, after examining later published reports and newer finds and reexamining the older finds made during Allen’s career, I now conclude that these groupings are, in a sense, artificial—particularly for dogs from the southwestern United States. In fact, the groups actually grade into one another in size, form, and amount of morphological variation, if a large enough collection is examined. The result is a more or less single mongrel group of southwestern Indian dogs, although the entire range of size and form variation is not found in every archaeological site. It is still possible to find representatives of only a part of the spectrum—either small, short-faced or long-faced dogs, or large Pueblo Indian dogs—at specified sites. The overlap is at the extremes of each of these size groups and is quite logical, because the prehistoric Indian dogs were hardly registered American Kennel Club breeds but were, instead, free-breeding, socializing mongrels.

It is, of course, quite simple to pick out representative animals of these differing groups from collections of excavated canids if there are enough individuals assembled from a large number of archaeological sites. The Basketmaker and early Pueblo Indian dogs of the Southwest and those from White’s Mound, Georgia, as well as the dogs from the shell heaps of Kentucky or Alabama, all show a close similarity in size and form. But these animals were quite advanced domestic dogs, and most were of a comparatively small size, although—as stated earlier—there are exceptions to this rule (for example, both small and large forms of dogs are found in the same strati-graphic level at the Jaguar Cave site).

Many early and later Pueblo Indian sites have yielded dog remains. In particular, reference is made to immature puppies as well as adults found in levels dating from about A.D. 1125 to 1175 at Antelope House in Canyon de Chelly, Arizona (Kelley 1975). At the Grasshopper Pueblo, built during the fourteenth century in east central Arizona, a number of small domestic dogs were recovered during 16 years of excavation by the University of Arizona Archaeological Field School (Olsen 1980). A unique find of an immature gray wolf, C. lupus, consisting of a right premaxilla and deciduous dentition, was also recovered from one of the rooms of this pueblo. This fragment of a wolf pup suggests that it may have been kept as a pet, perhaps with a view toward taming, although this is only speculation.

It is the case, however, that nearly every Pueblo excavation in the southwestern United States has produced some evidence of the domestication of the dog. These sites date from the eleventh through the fourteenth centuries and range in size from Cliff Palace Ruin in Mesa Verde National Park, Colorado (with its multistoried masonry construction of many rooms and great stone towers), to the more modest pueblos at Keetseel and Betatakin in the area of Monument Valley, Arizona. Tree-ring dates for Keetseel range between A.D. 1274 and A.D. 1284; those for Betatakin are between A.D. 1260 and A.D. 1277. By the time of these late dates, the domestic dogs were well advanced and were morphologically the same as present-day dogs.

As of this writing, however, we have yet to recover the ancestral forms of C. familiaris that would help to bridge the gap between the small, late Paleolithic wolves and the early Neolithic dogs from Asia and from Jaguar Cave, Idaho, on the one hand, and, on the other hand, the gap between the Jaguar Cave dogs and the well-known series of dogs of the southwestern United States that date from some 2,000 years ago and are still in existence today.