word _penetrable_, by
the addition of -ty. It is the Latin word _penetrabilitas_ imported.

_In derived words all the parts must belong to one and the same language_,
or, changing the expression, _every derived word must have a possible form
in the language from which it is taken_. Such is the rule against
hybridism.

s. 93. A true word sometimes takes the appearance of a hybrid without
really being so. The -icle, in _icicle_, is apparently the same as the
-icle in _radicle_. Now, as _ice_ is Gothic, and -icle classical, hybridism
is simulated. _Icicle_, however, is not a derivative but a compound; its
parts being _is_ and _gicel_, both Anglo-Saxon words.[39]

s. 94. _On incompletion of the radical._--Let there be in a given language
a series of roots ending in -t, as _saemat_. Let a euphonic influence eject
the -t, as often as the word occurs in the nominative case. Let the
nominative case be erroneously considered to represent the root, or
radical, of the word. Let a derivative word be formed accordingly, i.e., on
the notion that the nominative form and the radical form coincide. Such a
derivative will exhibit only a part of the root; in other words, the
radical will be incomplete.

Now all this is what actually takes place in words like _haemo-ptysis_
(_spitting of blood_), _sema-phore_ (_a sort of telegraph_). The Greek
imparisyllabics eject a part of the root in the nominative case; the
radical forms being _haemat-_ and _saemat-_, not _haem-_and _saem-_.

Incompletion of the radical is one of the commonest causes of words being
coined faultily. It must not, however, be concealed, that even in the
classical writers, we have in words like [Greek: distomos] examples of
incompletion of the radical.

* * * * *

s. 95. The preceding chapters have paved the way for a distinction between
the _historical_ analysis of a language, and the _logical_ analysis of one.

Let the present language of England (for illustration's sake only) consist
of 40,000 words. Of these let 30,000 be Anglo-Saxon, 5,000 Anglo-Norman,
100 Celtic, 10 Latin of the first, 20 Latin of the second, and 30 Latin of
the third period, 50 Scandinavian, and the rest miscellaneous. In this case
the language is considered according to the historical origin of the words
that compose it, and the analysis is an historical analysis.

But it is very evident that the English, or any other language, is capable
of being contempla