ted. The languages are very unequally
developed in their several parts. Low gender systems appear with high
tense systems, highly evolved case systems with slightly developed mode
systems; and there is scarcely any one of these languages, so far as
they have been studied, which does not exhibit archaic devices in its
grammar.

The author has delayed the present publication somewhat, expecting to
supplement it with another paper on the characteristics of those
languages which have been most fully recorded, but such supplementary
paper has already grown too large for this place and is yet unfinished,
while the necessity for speedy publication of the present results seems
to be imperative. The needs of the Bureau of Ethnology, in directing the
work of the linguists employed in it, and especially in securing and
organizing the labor of a large body of collaborators throughout the
country, call for this publication at the present time.

In arranging the scheme of linguistic families the author has proceeded
very conservatively. Again and again languages have been thrown together
as constituting one family and afterwards have been separated, while
other languages at first deemed unrelated have ultimately been combined
in one stock. Notwithstanding all this care, there remain a number of
doubtful cases. For example, Buschmann has thrown the Shoshonean and
Nahuatlan families into one. Now the Shoshonean languages are those best
known to the author, and with some of them he has a tolerable speaking
acquaintance. The evidence brought forward by Buschmann and others seems
to be doubtful. A part is derived from jargon words, another part from
adventitious similarities, while some facts seem to give warrant to the
conclusion that they should be considered as one stock, but the author
prefers, under the present state of knowledge, to hold them apart and
await further evidence, being inclined to the opinion that the peoples
speaking these languages have borrowed some part of their vocabularies
from one another.

After considering the subject with such materials as are on hand, this
general conclusion has been reached: That borrowed materials exist in
all the languages; and that some of these borrowed materials can be
traced to original sources, while the larger part of such acquisitions
can not be thus relegated to known families. In fact, it is believed
that the existing languages, great in number though they are, give
evidence of