sound-changes which one might imagine as happening in the future. A
language can tolerate only a certain number of ambiguities arising
from words of the same sound having different significations, and
therefore the extent to which a language has utilized some phonetic
distinction to keep words apart, has some influence in determining the
direction of its sound-changes. In French, and still more in English,
it is easy to enumerate long lists of pairs of words differing from
each other only by the presence or absence of voice in the last
sound; therefore final _b_ and _p_, _d_ and _t_, _g_ and _k_, are kept
rigidly apart; in German, on the other hand, there are very few
such pairs, and thus nothing counterbalances the natural tendency to
unvoice final consonants.'

[Footnote 12: _A Modern English Grammar on Historical Principles_,
by Otto Jespersen, Heidelberg, 1909. Streitberg's _Germanische
Bibliothek_, vol. i, p. 441.]

3. _That homophones are self-destructive and tend to become obsolete._

For the contrary contention, namely, that homophones do _not_
destroy themselves, there is prima facie evidence in the long list of
survivors, and in the fact that a vast number of words which have not
this disadvantage are equally gone out of use.

[Sidenote: Causes of obsolescence.]

Words fall out of use for other reasons than homophony, therefore one
cannot in any one case assume that ambiguity of meaning was the active
cause: indeed the mere familiarity of the sound might prolong a word's
life; and homophones are themselves frequently made just in this way,