æsthetic considerations are generally at one; and this blank statement
must here suffice, for the principle could not be briefly dealt
with: but it follows from it that the proper æsthetic objections to
homophones are never clearly separable from the scientific. I submit
the following considerations. Any one who seriously attempts to write
well-sounding English will be aware how delicately sensitive our ear
is to the repetition of sounds. He will often have found it necessary
to change some unimportant word because its accented vowel recalled
and jarred with another which was perhaps as far as two or three lines
removed from it: nor does there seem to be any rule for this, since
apparently similar repetitions do not always offend, and may even be
agreeable. The relation of the sound to the meaning is indefinable,
but in homophones it is blatant; for instance the common expression
_It is well_ could not be used in a paragraph where the word well (=
well-spring) had occurred. Now, this being so, it is very inconvenient
to find the omnipresent words _no_ and _know_ excluding each other:
and the same is true of _sea_ and _see_; if you are writing of the
_sea_ then the verb _to see_ is forbidden, or at least needs some
handling.

I see the deep's untrampled floor
With green and purple seaweeds strewn:

here _seaweeds_ is risky, but _I see the sea's untrampled floor_ would
have been impossible: even the familiar

The sea saw that and fled

is almost comical, especially because 'sea saw' has a most
compromising joint-tenant in the children's rocking game