Word count incorrect?

I’ve just compared the word counts produced by DEVONthink, Mellel and Nisus Writer Express. DEVONthink reckons a 280-word document (counted by hand) is only 260, while the others are far closer. Anyone else having the same problem?

A minor bug of the word count has been fixed in the latest release and hopefully the numbers are now accurate again.

It still seems to give consistently lower word counts than other programs (and, indeed, than WordService’s Statistics service). I think the problem is the way it handles figures. I’m using version 1.8.1b

Contrary to WordService, DEVONthink’s word count ignores “words” containing no alphanumerical characters, e.g. the string “word1 … word2 -#- word3” contains three words. WordService on the other hand scans only for white spaces and therefore says there are five words. But probably DT is more accurate, isn’t it?

Ah! Thanks for that. It also looks like DEVONthink’s word count ignores single digits (so that “Radio 4”, for example, counts as one word. I suppose that overall DT’s approach is the sensible one: it gives a fair impression of how long a piece is. The problem is that most other programs (Word, Nisus etc) seem to take the quick-and-dirty count-the-spaces approach. It’s confusing when you switch between programs, or when a publication that you’re writing for gives you a target word count based on a different approach. I guess the best solution is to check DT’s wordcount with the one produced by the Statistics service and sort of average them out.

Single characters/digits are not counted (this is quite often helpful but the english words "a" and "I" are therefore not counted too).

Maybe not quite on topic here…
If you want to know how (some) editors arrive at word counts, see
betweenborders.com/mss/index.html

About a third down the page is a paragraph on Word Count.

I don’t know why some editors are using word counts - e.g. I’ve been working as a journalist in the past and all editors specified the length of an article as a character count as this value is usually quite accurate, does not depend on word counting algorithms or on the average length of a word.

Because the editor usually knows which layout (e.g. number of pages/columns, size of columns, used font/size etc.) will be used and therefore it’s easy to calculate the maximal character count.

I like the idea of character counts - in principle. But in more than 15 years writing and commissioning for British newspapers and magazines (features, news, finance, arts), I’ve never used anything other than word counts, or even heard of anyone using them. I’ve had dealings with American writers and publishers as well. From what I recall, they’re all word-based as well. Were you working for German publications? Perhaps the practice is different for them.

Yes, I’ve been writing for German magazines and seems the practice is different over here (maybe because the length of German words varies a lot?).