The land surveys I work with were done in 1811. Three out of four surveyors used the same sampling methods. But one of them--was he confused? too innovative?--did not collect one of two types of data. Specifically, what surveyors were expected to do was to find the corner of the property lots and then locate the closest tree to that corner. They would write down the species of that tree, its diameter, and where it stood in relation to the corner (distance and compass direction). Then they would also blaze the tree with the lot numbers (carve the numbers into the bark). This information was collected so that whoever bought the property could then go out and identify the boundaries of their land. Unless the tree was struck by lightning and went up in a blaze, this was a fairly reliable way of keeping track of property corners for at least a couple of decades. And it helps forest scientists 200 years later.
But in the surveys I'm using, one surveyor did not record this information at all. Instead he had his team cut posts and set them in the ground at the lot corners. This sounds like a LOT more work than blazing a tree and taking down some notes. Moreover, the posts that he set were of ironwood--a small tree that is hard to cut but rots quickly. Those posts probably didn't stand for even a decade.
This is a common type of problem with historical data. It's full of gaps. It's not entirely reliable. It can't be checked! Or, in the case of historical records, it may have been gathered in a way that is difficult to reconcile with contemporary measures.
I read of an interesting example of using historical records which was recently published in the Journal of Geophysical Research. Woodworth (et al) made use of tidal gauge measurements in the Falkland Islands collected in the mid-19th century by the explorer James Clark Ross. They correlate these with other historical measurements of sea level and with contemporary measures based on satellite altimetry in order to construct a more long-term record of change in sea level.
What this historical data accomplishes is to show that the rate of sea level rise has been accelerating. In historical ecology, too, what the vegetation data show clearly is that there has been rapid change in the last 200 years but only slow changes in forest composition before that.
Historical data is gappy, but it's often good enough to demonstrate a key point.