This has been an interesting discussion with some informed answers.
The reason that I posed the question in the first place was in response to the many opinions expressed over the years that the Walker log was far more accurate than any through hull
type log, and I wondered about that.
By careful calibration vs GPS one can get a modern log to be fairly accurate, and if they are cleaned regularly the accuracy is maintained over time. Having done this, I have often then compared day's runs (GPS) with miles logged. Sometimes they agree pretty well, most times there are significant differences. These differences may stem from currents being present, from poor steering
or from instrument inaccuracy... possibly other things that I've not thought of.
I suspect that the primary cause is indeed varying currents encountered at sea. These currents do have a lot of fine structure... little whorls and eddies that, in the days before GPS, were undetectable by mariners. One has only to sit and watch the differential between GPS speed and knotmeter
speed as one proceeds across the sea to realize this. Or, have a look at the charts
published daily by various government
services showing the structure of important coastal currents such as the East Australia current
or the Gulf Stream
. They are very complex and change daily... those broad arrows shown on the pilot charts
are pretty misleading!
If one is using the log reading to establish an EP as an entry point for celestial nav calculations these inaccuracies are unimportant. If using for DR in poor visibility and a hazardous piloting situation, then they become more of a factor. We don't do that sort of thing much nowadays, so no one thinks about the issue!
But I was curious... and I thank all who contributed information.