The general standard to calculate total gross words, used in the typing contests and typing textbooks was to take the total stroke count (letters, characters, spaces) and divide by five. You would then divide by minutes to get the gross words per minute.

To get the net total, you would take the gross word total, then then deduct the penalty, which varied: the strictest was the 10 word per error penalty used in the Kimball's contests at the time of Albert Tangora's record (1923); the more lenient was deducting the actual words with errors from the gross total. After you calculate the total net word count, you divide by minutes to get net WPM.

M. Höhne, SoucekFan and Uwe, thank you for your thoughts.

For speed contests and newspapers then 5 characters per word seems to be the standard. I am surprised that includes spaces but the Wikipedia article seems to confirm this. Since we are counting strokes there may be a slight bias in favor or texts with short words. Spaces are almost guaranteed to be error free - it's hard to miss the space bar!

Wikipedia Words per Minute

The more I think about typing technique the closer it seems to piano technique. A pianist for example mentally plays phrases for the most part rather than individual notes, and a good typist types words and not individual letters. Common words become very well practiced typing phrases, as we all know, and an unfamiliar word can really slow you down. Entire English phrases may become practiced typing phrases with repetition.

The earlier penalty for errors (10 words/error) seems sensible from a business standpoint and the latter yet another example of the urge to inflate grades and measure screens diagonally. If I hire someone who can type 100 WPM raw with an average of five errors that is not the economic equivalent of one who can type 95 WPM error free! The documents must either go out with multiple errors or time must be taken to correct them, and if they must be corrected a steady and accurate 50 WPM starts to sound pretty good. This makes the 100 WPM - 10 x 5 = 50 WPM equivalence seem like a reasonable ballpark figure, though the real relative value would depend on context: if the goal is maximum rate of information transmission the sloppy 100 WPM fellow is putting a lot more of that stuff on paper and the redundancy of English is going to mean almost no loss of meaning, whereas if the material must be perfect then 50 WPM equivalent is possibly too generous.

Which error adjustment is closer to real economic value depends in the context after all. If I were running a typing contest I would offer a prize in both divisions - weighted for information throughput and under the demand of near total letter accuracy. In my private typing contests I grade myself on information throughput. Come to think of it when I played the piano my standard was pretty much the same. 

SoucekFan wrote:

The error deduction is an important factor, as just typing fast is not nearly as impressive.

I did show an awareness of that and discussed when the stricter and more lenient measures may have been an appropriate figure of merit. Please see my third paragraph instant. 

SoucekFan wrote:

Also, the fact that the professionals contests not only typed at these speeds with high accuracy, but for 1-hour periods is mind blowing. The Amateurs did 30 min, and the Novices 15 minutes. According to the Kimball's contest rules: from 1906-1918, there was a 5-word deduction for errors. From 1918 to sometime in the early 30's it was a 10-word deduction, and then sometime prior to 1934 the 10-word error was abolished in favor of giving credit for every perfect word. Other contests may have varied, but Kimball's was the big Underwood sponsored event. You could actually have an error on spacing, as well as margins and improper shifting. The accuracy and endurance of the typing champions was as amazing as their speed.

I had however overlooked endurance, and being to keep up that kind of speed and accuracy for an hour is out of this world. That's like playing the Rachmaninoff 2nd and 3rd piano concertos back to back but leaving out the rests and slow sections! I freely admit that my endurance is in the novice range and I aspire to the exalted level of amateur. I hope I have at least crawled out of the dilettante or "pre-K" group. 

I would not however bother with aspiration on the more rigorous error correction scales as this would not suit my purpose. I am not typing business correspondence and I am more interested in the speed of getting ideas onto paper, relying on the redundancy of English to make the meaning clear when a few bits get misplaced.