Re: Types of numerals; bases in natlangs.

AJ = Andreas Johansson
TM = Tristan McLeay
VP = Veritosproject

AJ> There's also Mo = megaoctet. But I seem to see MiB, TiB, etc with some
AJ> frequency.

TM> "Megoctet", I think, isn't -a dropped before o-? (Also "megohm".)

Final vowels in the SI prefixes in general, in English at least since
the spoken prefixes vary by language, are dropped before a vowel when
the bivocalic pronunciation is considered "awkward" - thus, 1/1000th
of an ohm is milliohm, not *millohm, even though 1000000 ohms is a
megohm, not *megaohm.  A notable exception to this rule is the cgs
unit "erg", which sprouts a leading L when the prefix ends in a vowel:
megalerg, not *megerg.

Since the word "octet" starts with a different vowel from "ohm" (/O/
or /a/, as opposed to /o/), its awkwardness is independent, despite
the same spelling.  It could go either way; I find "megaoctet" rather
less awkward than "megaohm"; the final /@/ of "mega' tends to
disappear entirely in the latter, but not in the former.

VP> MiB, KiB are "mibibyte" and "kibibyte".  Technically, unlike what you
VP> have heard, a kilobyte and megabyte are really exactly 1000 and
VP> 1000000 units.  A kibibyte is 1024 bytes and a mibibyte is 1024
VP> kibibytes.

A clear case of prescription flying in the face of usage.  Almost
nobody uses the binary prefixes, but still use "kilobyte" to refer to
1,024 bytes.  In their seven years of existence, the binary prefixes
have won some converts, but their use is still very much in the
minority.  And the International Electrotechnical Commission's
recommendation for their use is not really enough to allow a blanket
statement that the use of the older prefixes for powers of 2 is
"technically" incorrect.  Beware the sweeping generalization. :)
--
Mark J. Reed <markjreed@...>