Regnault's tables of the thermodynamic properties of steam and other data w
ere standard during much of the 19th century. To calculate them from the ex
perimental measurements, he used a graphical method to distinguish between
random and systematic errors, when both were so small that he had to plot p
oints with an accuracy of one-hundredth of a millimetre. He did this on a l
arge copper plate and did the curve fitting on the plate. This episode is s
et against the development of the drawing of experimental graphs from the m
id-18th century onwards.