THE SCIENCES OF LANGUAGE AND OF ETHNOGRAPHY. 111 
of others. In other words, in addition to the mere elementary 
acquisition of knowledge, you have to cultivate a sympathetic 
attitude; and here, again, is one of the proofs of a truth which 
my experience has taught me, that, however great knowledge 
may be, sympathy is greater, for sympathy enables us to fit the 
key which is given by knowledge. Gestures also elicit a response 
in dealing, for instance, with numerals, where we are facilitated 
by the fingers of the hand. Of course, one is occasionally 
stopped by a savage who cannot go, or is supposed not to 
be able to go, beyond two, or beyond tive. 
I take it that in the majority of cases of that kind, a good 
deal of our misconception with regard to the difficulty of 
the inquiry lies in ourselves—that ideas of multitude 
connected with the peculiar customs of the race that have 
yet to be ascertained, are at the bottom of the inability of 
that race to follow our numeration. Tor instance, we go up 
to ten, and in order to elicit a name for eleven, we say “ one, 
ten”; if the man laughs, change the order, and say “ ten, 
one’; the chances are that the savage will imstinctively rejoin 
“ten and one,” and we then get the conjunction. Putting 
the fingers of both hands together may mean ‘ multitude,” 
“ alliance,” or “ enmity,” according as the customs of thé 
race are interpreted by that gesture. 
Tam reminded of this particular instance in my experience, 
because I referred to it in a discussion on an admirable paper 
on the Kafirs of the Hindukush by the eminent Dr. Bellew, 
who, I hoped, would have been present this evening. If you do 
not take custom along witha “rule,” and do not try to explain 
the so-called rule by either historical events or some custom 
of the race, you make language a matter entirely of memory, 
and as memory is one of the faculties that suffers most 
from advancing age, or from modes of living and various other 
circumstances, the moment that memory is impaired your 
linguistic knowledge must suffer,—you, therefore, should make 
language a matter of judgment and of associations. If you 
do not do that, however great your linguistic knowledge or 
scholarship, you must eventually fail in doing justice to the 
subject or to those with whom you are dealing. 
The same principle applies as much to a highly civilised 
language like Arabic, one of the most important languages 
in the way of expressing the multifarious processes of human 
thought and action, as to the remnant of the pre-historic 
Hunza language with which I am going to deal to- 
night, and which throws unexpected light on the science 
of language. 
Let us first take Arabic and the misconceptions of 
