Although I was initially put off by the somewhat negative reviews of the book, and the reports of copious numbers of mistakes, having now obtained a copy and gone through it thoroughly, I think it's fair to say that the underlying data is sound.

Everything seems to make sense - the vast majority of records start charting in the month they were released, and there is a reason for most of those that don't, e.g. a revival in the song due to a new version by say Bing Crosby being released. As an example, there's a group of Brunswick records in Jan 40 that are charting long after their release date, which I initially was sceptical of, until I came across Newspaper adverts that Brunswick had placed around Christmas, listing these records and boasting about their sales figures in the US ("Hundreds of thousands can't be wrong"!). Obviously everyone subsequently went out and bought them (or at least publishers/wholesalers ordered them in).

The main criticism I have is the identification of the "A" side of a record. This seems to have been done fairly haphazardly, and I don't believe there is any way they could have known for certain which side was more popular. The concept of A and B sides didn't really exist at the time (the Decca group did label their records with A and B, but I believe this was simply based off the matrix numbers), and particularly during the war, when the number of record releases was limited, pretty much every record would have had a popular song on each side.

As for the mistakes in the book, there are certainly hundreds of typos, mis-attributed artists, wrong titles, catalogue numbers etc, which is very disappointing (and frustrating) but I don't think should be taken as an indicator of the quality of the underlying data. I suspect the number of errors was not helped by the fact that Colin Brown died before the book was completed, as I don't think the other authors involved were particularly au fait with the music of the period.

So far I have found only 20 records (out of the 2325 listed) that are more serious errors, i.e. they are credited as charting before (or long after) they were released. There may be a few more, the release dates of which I am still trying to confirm, but not many. Most of these are out by only a few months, and some in fact are credited as having a second charting period when they were actually released. There are 5, however, that are out by exactly a year, these being:

1944_155 HMV BD-1119 - Pat McCormac - Silent Night, Holy Night (charted Dec 45, released Dec 46)
1951_081 Parl. R-3290 - Five Smith Brothers - Dearie (charted Jun 51, released Jun 50)
1951_082 Parl. R-3292 - Humphrey Lyttelton - Ice Cream (charted Jun 51, released Jun 50)
1952_087 Col. DB-3308 - Doris Day - If I Were A Bell (charted Jul 52, released Jul 53)
1952_099 Col. DB-3306 - Ray Burns - Eternally (charted Jul 52, released Jul 53).

I don't think that this is a co-incidence - I suspect what has happened is that the sales for these have been attributed to the wrong year at some early point in the process. Funnily enough, the folks over at the "Real Chart" project also have these same errors for 1951 and 1952 (see here). This is despite their assertion that they are using a different dataset from the Missing Charts book (actual sales rather than orders), but given the similarity in errors I think this claim is questionable...