decimating DECIMATE

Philip (Flip) Kromer flip at MRFLIP.COM
Sun Jan 6 11:31:48 UTC 2008

I'm sorry to wander in near the tail end of this discussion, but as a
computer geek (and not at all a word expert) I'm curious why none of
the dictionaries I can access (OED, M-W, Bartleby, MSN) mention the
very common
signal-processing sense of decimation: 'divide a sample into, then
eliminate a fraction of, many small slices". In addition to the
wikipedia entry, google shows about a million hits for
(decimation OR decimate) (sampling OR signal OR MPEG)

It appears to be an established term of art by 1970, its first mention
in the patent database:
"Each of decimators 310-1 and 311-1 is operative, as is well known
in the art, to decimate the 10 kHz sampling rate by a factor of 25
so that the sample signal output of each of the decimated samples
occurs at a 400 Hz rate."
(You gotta love patent application prose.)

I think (again, as a non-expert) this weakens the "deci- has a strong
one-tenth implication", since in practice the sampling rate reduction
would be chosen as 10% or 90% only by coincidence. This context for
the word -- directily implying a fractional reduction of the original
total -- would seem to act towards a flavor of "decimal" and
"decigram", yet engineers feel no shame in decimating by 2, or 25, or
25 / 29.97ths.

Also of note: among the
six pre-1918 appearances of 'decimate' in the patent library, there
are two uses as "devastate", one as "divide into tenths or decimally",
and three 'kill off a fraction of' (no more than one in the 'reduce by
90%' sense).


> Date: Thu, 3 Jan 2008 12:00:10 EST
> Reply-To: American Dialect Society <ADS-L at LISTSERV.UGA.EDU>
> Sender: American Dialect Society Mailing List <ADS-L at>
> From: RonButters at AOL.COM
> Subject: decimating DECIMATE

> I agree that most of the arguments about DECIMATE are ignorant and peevish,
> but I don't think it is entirely a matter of "the etymological fallacy." At least as
> important is the relative transparency of "DECI" (as in "decimal," etc.), which
> has a sort of independent morphosemantic existence that allows (causes?)
> folks to associate the word with the meaning 'ten'.

The American Dialect Society -

More information about the Ads-l mailing list