This is that which is vaguely known as El Gran Chaco, or the Great Hunting Ground. It lies in northern Argentina and eastern Bolivia, between latitude 18° and 32° south, and longitude 58° and 66° west from Greenwich. It covers an area about as great in extent as from Pittsburg to the Mississippi and from Chicago to the Gulf of Mexico.1

On the east, the valley of the Rio Parana and Rio Paraguay, which are the extensions of the Rio de la Plata, and on the west the lofty elevations of the Andes, are its well-marked boundaries. Between them the surface is usually level and intersected by numerous streams, the three most important of which, the Pilcomayo, the Vermejo and the Salado, flow from the Andes southeastward in almost parallel lines.

The climate is hot and the vegetation tropical. During the rainy season the flat, grassy lands are transformed into shallow lakes, while near the watercourses rise dense and lofty forests. In the north are arid and sterile highlands.

Except by the water-ways it is almost impossible to traverse the country, and for that reason extensive tracts of it are still unexplored.

The native tribes who inhabited this region have always been in the lowest stages of culture, depending on hunting and fishing for their subsistence, without settled abodes, migratory and in ceaseless warfare with each other. The self-sacrificing efforts of the Jesuit and Franciscan missionaries have at times succeeded in gathering a few hundred together about some mission, only to be dispersed again on some slight cause. Thus, some years ago, in the middle of the night, the whole of the tribe of Penoquiquias, which had been converted and induced to take up a fixed abode, suddenly disappeared, and were never seen again (Cardus, i, p. 272).

RECENT CONTRIBUTIONS TO THE LINGUISTICS OF THE CHACO.

In my volume on the linguistic classification of the American. race, published in 1891, I divided the tribes of the Chaco into four principal linguistic stocks, the Guaycuru, the Mataco, the Vilela

1 Boggiani (i, p. 10) puts the maximum length of the Chaco at 830 geographical miles, and its greatest width at 360 miles.

For this and other references in the text see the Bibliographic Note at the close of this article.

and the Payagua. Since then a number of contributions to the subject and several ethnographic maps of parts of the region have been furnished by J. B. Ambrosetti, J. de Brettes, Guido Boggiani, . Dr. G. A. Colini, Giovanni Pelleschi, Samuel A. Lafone Quevedo, and others, and some unpublished manuscripts of early date have seen the light in print, the titles of which are given in the bibliographic note at the close of this paper. My present intention is to offer a summary of their results in the linguistic ethnography of the Chaco region, as depicted in the revised chart which I have prepared, and to present some suggestions for the correct classification of tribes of still uncertain affinities.

The writer to whom we owe most is Mr. S. A. Lafone Quevedo. He has edited the manuscripts of Tavolini, Brigniel, Barcena and others, and contributed numerous articles of his own, and deserves the highest credit for his zeal and accuracy.

He entertains, however, certain linguistic theories which would with difficulty find general acceptance, and which expose some of his reasonings to serious question.

Thus, he maintains that there is a fundamental difference between what he calls the "Atlantic" type of languages and the "Pacific" or "Andean," based upon the trait that the latter suffix the pronominal particles while the former prefix them.

This, as a mere matter of placement, is of minor importance. For example, the Latin has the pronominal indications suffixed to the verbal root; but the neo-Latin tongues without exception prefix them. In America, the dialects of the Zoque-Mixe stock differ entirely among themselves in this respect, though closely allied in others.2

He himself acknowledges (x, p. 23) that the dialects of the Guaycuru stock are by no means a unit in this feature, some prefixing and others suffixing the pronominal particles.

The American Race. A Linguistic Classification and Ethnographic Description of the Native Tribes of North and South America, pp. 392 (New York, 1891).

2 Raoul de la Grasserie, Langue Zoque et Langue Mixe, Introduction (Paris, 1898).

Yet in some passages (xi, p. xliii) he dismisses the criterion of the affixes, and in maintaining the affinity of Quechua and Guaycuru says their contrast in this respect is unimportant!

I do not question the general value of pronominal pre-position and post-position; but it is not sufficiently fundamental to be adopted as a single criterion for classification.1

Another feature to which Mr. Lafone Quevedo has given close study is the permutation of sounds in these tongues. Undoubtedly he has here shown regular and frequent changes between the dialects. But there will be few to follow him in such an equation as:

With such liberty, any two words could be brought into genetic relation.

This laxity of method naturally leads him to assert linguistic affinities between all stocks; these he claims the Guaycuru has with the Guarani, the Mataco with the Carib, and all with the Arawack (L. Q., ii, pp. 56, 58).

Such conclusions are regretable, and it were to be desired that students of American languages should be as cautious in asserting analogies as are the leading scholars in the Aryan and Semitic fields.

THE MATACO LINGUISTIC STOCK.

The linguistic study of the tribes of this vast area has had the usual effect of constantly reducing the number of its linguistic stocks by recognizing as dialects what earlier observers believed to be independent languages. To this result, I shall also contribute somewhat in the present article.

It has been long recognized that most of the Chaco region was occupied at the discovery by two great groups of related idioms.

One of these was central, extending in unbroken continuity from the river Paraguay to the foothills of the Andes, and from S. lat. 21° to 26°. This was the Mataco stock, so called from its central and principal tribe." It is noticeable that all its members

1 Prof. Friederich Müller observes that while there are a number of suffix-languages, there is, in fact, no example of a true prefix-language, "da neben ihr immer die Suffix-Bildung zur Anwendung kommt" (Grundriss der Sprachwis senschaft, Bd. i,p. 129). This iliustrates how uncertain such a criterion must be. Prof. Hovelacque remarks that linguistically the position of the pronominal affix "n'a aucune importance" (La Linguistique, p. 174.)

2 The Matacos refer to themselves as vicquii viri, men.

reside in contiguity, as if driven together by outer pressure on all sides. This would give probability to the opinion that they are the oldest surviving inhabitants of the Chaco.

The dialectic variations in the stock may be seen in the following comparisons:1

The tribe called by de Brettes the Aksseks must be members of the Mataco stock. They dwelt on the Bolivian frontier, extending in a northwesterly direction from the Rio Pilcomayo to the confines of the Samucus.

Lafone Quevedo is of the opinion that the Mataco is a jargon, owing its lexicon to one stock and its grammar to another (Pelleschi, ii, p. 14). This is not the impression that it makes upon me. I rather agree with Father Remedi (in Lafone Quevedo, v, p. 25)

1 I have selected the same words which form the brief comparative vocabularies in my American Race.

2 Mallat de Bassilan, L' Amérique Inconnue, pp. 11, 37. They spoke neither Guana (Ennima), Guarani or Chamacoco.

that it is of a very ancient type, and, apart from a certain number of borrowed words, is a wholly independent stock.

THE GUAYCURU STOCK.

The second stock has received the name Guaycuru, a Tupi word meaning "fast runners."'1 Mr. Lafone Quevedo does not acknowledge a fundamental distinction between this and the preceding group of tongues, but the evidence seems to me insufficient to blend them in one.

They almost surrounded the Matacos on three sides, the south, east and north, and extended on the eastern bank of the river Paraguay as far as S. lat. 19° 30' into the Brazilian province of Mato Grosso. In the south they roved as far as to 33° lat. south, where they were in contact with the Pampean tribes.

One of their most redoubtable members were the Tobas, called by the Spanish Frentones or Frontones, from their habit of shaving the forehead. In modern times they have occupied the shores of the lower Pilcomayo, and have been reported at various localities along its banks quite up to its rapids (Cardus, i; Thouar, i). The Pilagas are a closely related horde.

The large majority of this stock lived west of the great river; but in the south the Abipones and Mepones were found in the last century east of the river about lat. 28°; and in the north all the left bank, from Fort Coimbra to Puerto Casado, was peopled by tribes of the Guaycuru stock, locally known as Quetiadegodis or Uettiadiu, and Eyiguayegis or Eggiuágeg, the modern Mbayas and Caduveos (Boggiani, iv, p. 171).

The dialectic variations in this stock may be seen in the subjoined table.

1 Lafone Quevedo considers this word a diminutive from the root ai, rogue; but I think that von Martius is right in considering it composed of atá or guatá, to go; curitei, quickly; uara, men. Some writers have objected to the use of this collective name for the stock on the ground that it is a common noun, and does not apply to a single nation. The same objection would be applicable to many nomina gentilia in common use (e. g. Aryans, Semites) and is therefore a needless criticism. There are reasons why it is not desirable to choose the name of a single tribe for the whole stock.

2" El grupo Mataco es una subclase mas del gran grupo Guaycuru." (Bol. Inst. Geog. Argentin, 1894, p. 518.)

3 M. Thouar (1, pp. 419–421) gives vocabularies of the Toba on the lower and on the upper Pilcomayo.