M47 was missed because Messier did a sign error during the reduction of positional data. He computed the position of this cluster from the differences to the star 2 Puppis (2 Navis in Messier's time), but mistook the sign of the right ascension difference. This fact was recognized by T.F. Morris of the RAS of Canada in 1959, after Oswald Thomas had identified it correctly as NGC 2422 in his 1934 book Astronomie. Previously, John Herschel had given the wrong position a number in his General Catalogue, and following him, J.L.E. Dreyer a NGC number, NGC 2478, although at that position there's no object at all.
M48 is not so obvious, but as (the same astronomer) T.F. Morris pointed out in 1959, the only object matching Messier's description in this celestial area is NGC 2548, which is now generally recognized as M48. It lies at the same right ascension, but almost exactly 5 degrees south of Messier's position. The reasons for this error will probably remain obscure unless Messier's lost observing books of this period should come to light one day.
M91 was much more difficult to reconstruct; finally, Messier had measured the position of this galaxy from the previously discovered M89, but thought he had used M58, as the amateur astronomer William C. Williams of Texas had found out and thus identified M91 with NGC 4548, now generally accepted and quite safe. Previously, it had been assumed that M91 might be the 12 mag galaxy NGC 4571, the nearest to Messier's position - unlikely but not totally impossible. Many sources also held the version that it was actually a comet that fooled Messier - even more unlikely with regard to the fact that Messier was the comet specialist, and Owen Gingerich had brought up the hypothesis that it might be a duplicate observation of M58.
M102 finally could not be cleared up with certainty up to now. At last, there are still two possibilities open: It may be a duplication of M101, as its discoverer Pierre Mechain believed when he wrote a letter to Bernoulli in Germany two years later, but on the other hand, its description in Messier's catalog (which was actually Mechain's description) matches well with NGC 5866. Moreover it may be that Charles Messier has observed this object when measuring the position of M102 which he wrote by hand into his personal copy of the catalog, but did a data reduction error again, plotting it exactly 5 degrees west (preceding) of its true position in right ascension. The present author has discussed this topic and thinks it depends on taste to believe which was erroneous: the observation or the letter, or if Messier's possible observation justifies the designation `M102' for this object.
To summarize: The four missing Messier objects were probably missed because of errata of Messier in data reduction, in detail one sign error (for M47), one mistaken comparison object (for M91) and one or, probably, two "grid" errors, i.e. positions which are exactly 5 degrees off (for M48 in declination, for the M102 candidate NGC 5866 in right ascension), which is the grid tick width in the charts he used (see e.g. this example) and can thus be explained by wrong looks or labels.
Last Modification: 25 Jan 1998, 14:50 MET