In 1836 Gallatin conferred a great boon upon linguistic students by classifying all the existing material relating to this subject. Even in the light of the knowledge of the present day his work is found to rest upon a sound basis. The material of Gallatin’s time, however, was too scanty to permit of more than an outline of the subject. Later writers have contributed to the work, and the names of Latham, Turner, Prichard, Buschmann, Hale, Gatschet, and others are connected with important classificatory results.

      The writer’s interest in linguistic work and the inception of a plan for a linguistic classification of Indian languages date back about 20 years, to a time when he was engaged in explorations in the West. Being brought into contact with many tribes, it was possible to collect a large amount of original material. Subsequently, when the Bureau of Ethnology was organized, this store was largely increased through the labors of others. Since then a very large body of literature published in Indian languages has been accumulated, and a great number of vocabularies have been gathered by the Bureau assistants and by collaborators in various parts of the country. The results of a study of all this material, and of much historical data, which necessarily enters largely into work of this character, appear in the accompanying map.

      The contributions to the subject during the last fifty years have been so important, and the additions to the material accessible to the student of Gallatin’s time have been so large, that much of the reproach which deservedly attached to American scholars because of the neglect of American linguistics has been removed. The field is a vast one, however, and the workers are comparatively few. Moreover, opportunities for collecting linguistic material are growing fewer day by day, as tribes are consolidated upon reservations, as they become civilized, and as the older Indians, who alone are skilled in their language, die, leaving, it may be, only a few imperfect vocabularies as a basis for future study. History has bequeathed to us the names of many tribes, which became extinct in early colonial times, of whose language not a hint is left and whose linguistic relations must ever remain unknown.

      It is vain to grieve over neglected opportunities unless their contemplation stimulates us to utilize those at hand. There are yet many gaps to be filled, even in so elementary a part of the study as the classification of the tribes by language. As to the detailed study of the different linguistic families, the mastery and analysis of the languages composing them, and their comparison with one another and with the languages of other families, only a beginning has been made.

      After the above statement it is hardly necessary to add that the accompanying map does not purport to represent final results. On the contrary, it is to be regarded as tentative, setting forth in visible form the results of investigation up to the present time, as a guide and aid to future effort.

      Each of the colors or patterns upon the map represents a distinct linguistic family, the total number of families contained in the whole area being fifty-eight. It is believed that the families of languages represented upon the map can not have sprung from a common source; they are as distinct from one another in their vocabularies and apparently in their origin as from the Aryan or the Scythian families. Unquestionably, future and more critical study will result in the fusion of some of these families. As the means for analysis and comparison accumulate, resemblances now hidden will be brought to light, and relationships hitherto unsuspected will be shown to exist. Such a result may be anticipated with the more certainty inasmuch as the present classification has been made upon a conservative plan. Where relationships between families are suspected, but can not be demonstrated by convincing evidence, it has been deemed wiser not to unite them, but to keep them apart until more material shall have accumulated and proof of a more convincing character shall have been brought forward. While some of the families indicated on the map may in future be united to other families, and the number thus be reduced, there seems to be no ground for the belief that the total of the linguistic families of this country will be materially diminished, at least under the present methods of linguistic analysis, for there is little reason to doubt that, as the result of investigation in the field, there will be discovered tribes speaking languages not classifiable under any of the present families; thus the decrease in the total by reason of consolidation may be compensated by a corresponding increase through discovery. It may even be possible that some of the similarities used in combining languages into families may, on further study, prove to be adventitious, and the number may be increased thereby. To which side the numerical balance will fall remains for the future to decide.

      As stated above, all the families occupy the same basis of dissimilarity from one another—i.e., none of them are related—and consequently no two of them are either more or less alike than any other two, except in so far as mere coincidences and borrowed material may be said to constitute likeness and relationship. Coincidences in the nature of superficial word resemblances are common in all languages of the world. No matter how widely separated geographically two families of languages may be, no matter how unlike their vocabularies, how distinct their origin, some words may always be found which appear upon superficial examination to indicate relationship. There is not a single Indian linguistic family, for instance, which does not contain words similar in sound, and more rarely similar in both sound and meaning, to words in English, Chinese, Hebrew, and other languages. Not only do such resemblances exist, but they have been discovered and pointed out, not as mere adventitious similarities, but as proof of genetic relationship. Borrowed linguistic material also appears in every family, tempting the unwary investigator into making false analogies and drawing erroneous conclusions. Neither coincidences nor borrowed material, however, can be properly regarded as evidence of cognation.

      While occupying the same plane of genetic dissimilarity, the families are by no means alike as regards either the extent of territory occupied, the number of tribes grouped under them respectively, or the number of languages and dialects of which they are composed. Some of them cover wide areas, whose dimensions are stated in terms of latitude and longitude rather than by miles. Others occupy so little space that the colors representing them are hardly discernible upon the map. Some of them contain but a single tribe; others are represented by scores of tribes. In the case of a few, the term “family” is commensurate with language, since there is but one language and no dialects. In the case of others, their tribes spoke several languages, so distinct from one another as to be for the most part mutually unintelligible, and the languages shade into many dialects more or less diverse.

      The map, designed primarily for the use of students who are engaged in investigating the Indians of the United States, was at first limited to this area; subsequently its scope was extended to include the whole of North America north of Mexico. Such an extension of its plan was, indeed, almost necessary, since a number of important families, largely represented in the United States, are yet more largely represented in the territory to the north, and no adequate conception of the size and relative importance of such families as the Algonquian, Siouan, Salishan, Athapascan, and others can be had without including extralimital territory.

      To the south, also, it happens that several linguistic stocks extend beyond the boundaries of the United States. Three families are, indeed, mainly extralimital in their position, viz: Yuman, the great body of the tribes of which family inhabited the peninsula of Lower California; Piman, which has only a small representation in southern Arizona; and the Coahuiltecan, which intrudes into southwestern Texas. The Athapascan family is represented in Arizona and New Mexico by the well known Apache and Navajo, the former of whom have gained a strong foothold in northern Mexico, while the Tañoan, a Pueblo family of the upper Rio Grande, has established a few pueblos lower down the river in Mexico. For the purpose of necessary comparison, therefore, the map is made to include all of North America north of Mexico, the entire peninsula of Lower California, and so