Dear Masashi,
The thing is that, unlike English (but like Spanish, French, Italian,
Portuguese, etc...), Latin names belong to one out of three genders:
masculine, feminine or neuter. It is not always easy to know what gender
things belongs to, because Latin use a system of five declensions where
genders may be mixed, but as a first approach you can asume that names
fininished in -a are feminine, in -o, -us are masculine and those finished
in -um or consonant tend (but not always) to be neuter.
Linnaeus, like all of that time scientist, speaks Latin very well. The
problem arise when a species is given a new generic name (that could belong
to another gender) but retains unchanged its specific name. This drives
directly to a grammatical error, but if (and then my question) ICZN accepts
the new name as valid, this error will be perpetuated unless there is a
rule that gives priority to grammar over nomenclature.
Hope this help you to understand the thing,
Best regards,
Alfonso Pina
Malaga
Spain