NavList:

Geoffrey Kolbe, you wrote:
"They did not have an absolute reference, such as a GPS receiver, to check on their workings and sightings as we do, so it is difficult to know just how well 19th century deep sea mariners knew their actual position."

There are actually several ways of doing this. The best are when the navigators report their latitude and longitude when near a known island or headland.

You continued:
"However, they also noted that they found 32 examples where the logs of two ships each recorded a sighting of other, and gave the position. The mean absolute difference  between pairs of reported positions was 0.22 degrees of latitude (with an SD of 0.315 degrees) and 0.54 degrees in longitude (with an SD of 0.667 degrees)"

According to the article, they say that these are cases where one vessel "spoke" another ship. And, yes, that's the slightly odd way that they would put it, for example, they might record, "Today we spoke the Bark California" and in general this was called "speaking other ships" which meant getting close enough to exchange information either by yelling at each other or by using messages on slates seen through telescopes, but surely less than about 500 feet apart. Point being, they didn't simple "sight" each other. These recorded differences would reflect the maximum typical differences, since vessels only asked for and recored position data when in doubt. Early in the century, even in the 1840s, doubt over longitude, especially, was quite common, and position data was frequently exchanged. By the end of the century, there were far fewer cases where position data was exchanged and recorded. But there's a much bigger problem: position as received from the "other" cannot be directly compared with the position aboard "our" vessel. The "speaking" event as recorded may have occurred at anytime during that day (and it's the position data at that time which would normally be noted, not updated in any way) while the position recorded in the logbook was almost universally the position at noon. It would not be unusual for two whaling vessels to separate by a dozen miles during the course of an afternoon...

I estimate from my logbook analyses that latitudes when "by obs" in the 19th century were generally accurate to two or three miles throughout the century (in the s.d. sense), much as they were in the early 20th century. Longitudes were never that good, but by the end of the century, nearly so. Early in the century, longitudes were naturally much worse. And of course it depended on the available tools. I've seen cases where the master of a vessel spoke another ship, got the other's longitude (rarely asking for the latitude since that wasn't a concern) and then ignored it saying, "His longitude was [some number] but he has no chronometer and knows nothing of lunars so I dismiss it."

Frank Reed