The apparatus that Michelson brought to the BIPM in 1892 to “determine the value of the metre in luminous wavelengths” has generated a vast family of instruments typically used for length measurement. Looking at them from this point of view teaches us a lot about the characteristics of the Michelson interferometer that make it specially convenient for Fourier transform spectroscopy (FTS). Michelson himself used the principle of the FTS to check the monochromaticity of various spectral lines. He was looking for a good standard of length, hence he did not care so much about the general problem of spectroscopy. Furthermore, he was limited to the parameter that he called the visibility of the fringes, i.e. the modulus of the Fourier transform of the spectral distribution, in fact the only parameter that the naked eye can roughly estimate, besides the fringe spacing. As he counted only up to 1212 fringes, and thereafter used a stepping process by “optical addition”, he did not worry about the optical defects that impair direct interferometric measurement of long distances. A few technological improvements have made the fringe-counting technique the current way of measuring length. It is shown that the analysis required for accurate length measurements with a Michelson interferometer is similar to the one required to make accurate FTS.