The matrix numbers are the numbers pressed into the "dead wax" close to the labels of the record:

Machine-stamped matrix number on first US issue of TMWSTW (To make it more easily visible, I had to change the contrast and brightness of the scan.)

The matrix numbers are not always easy to detect. Sometimes you have to hold the record under a bright light to see them.
Why could these numbers be important to record collectors? Well, they sometimes helps to identify different pressings of a record. Usually, the numbers are machine-stamped, but sometimes they are hand-scratched. But this is not always a reliable criterion to distinguish originals from counterfeits or re-issues, since especially in the USA the numbers are often hand-scratched - even on perfectly legal, original records. However, there are cases in which the matrix number is the decisive criterion. There is no general rule, though!
Moreover, when the matrix of a record is worn out, a new one is used. Usually, it has got a different number then. It may seem logical to give the first matrix used for pressing the A-side of a particular record the number "A1" (or something like that). Some collectors have turned this into a kind of religion: they try to collect records with the matrix number combination "A1/B1" or so. However, when several copies of a record are pressed simultaneously, then several matrices must be used. So right from the beginning, matrix "A1" might have been used at the same time as "A2".
Ultimately, matrix numbers might help to distinguish first pressings from later pressing, but there is no general rule concerning the number system.
By the way... if several matrices were made, who can tell which one the workers at the pressing plant decided to use first: "A1" or "A2"? So "A2" might even have been used before "A1".
To sum up: there are cases in which the matrix numbers are helpful to distinguish different issues of a record, but apart from these few cases I wouldn't care too much about them!

Early 1970s RCA Germany label

Early 1980s RCA Germany label

- the label colour
- the typeface of the text
- the exact spacing and position of the text
- additional information on the labels, like label codes ("LC...") on the labels of German LPs and records from neighbouring countries
- the copyright company listed on the labels (like "GEMA", "BIEM", etc. - often in a square box), and, if there is more than one such company listed, the order in which they are listed
- the details of the text running around the label margins
- the width of the space between that text and the label margins
- the label moulding (i.e. the way that grooves are pressed into the label) - which is why I also show white labels (of bootlegs, for example)
- and possibly other things I can't think of at the moment...

As a general rule, then, any date given on the record is the earliest possible date the record might have been released, but it might also have been released later.

So the information given on the records themselves may be helpful sometimes, but to actually determine the release date of a record, you have to confer further sources. Jarman, for example, studied the music magazines (etc.) of the time to determine the month of release. Not surprisingly, I didn't do that, but mainly relied on second-hand data (and occasionally on my own memory). My main sources were Jarman and Jarman & Stöcklin (for 7" singles up to 1981), Pegg, Thompson, Carr & Murray (for early US LPs), and the various books by Frifelt (for bootlegs). Online sources I have used include Bassman's David Bowie Page, the Illustrated db Discography, the David Bowie 7" Singles website, and even the various Amazon websites. See the "Books and Links" section for a more complete list.
If there was conflicting information, I have chosen the date that (for whatever reason) appeared to be most plausible to me. Very often this is pointed out in the "Collector's notes". If I couldn't find any information at all, I have just listed the record at the place where... it seemed plausible.
The result is not perfect, of course. But even Jarman, who probably did the most impeccable research into release dates, had to file some "exotic" releases under the release dates of the issues from "major" countries...

2. On counterfeits
Let's start from the following preliminary definition: counterfeits are unofficial imitations of officially released records. Some people consider only those items to be counterfeits that are actually meant to trick collectors into thinking that they are buying the real thing. Good examples of counterfeits that can easily fool unexperienced collectors are the counterfeit of the US mono Deram LP, the fake German TMWSTW round cover LP, and the 2014 issue of the UK TMWSTW dress cover LP. However, the restriction that a counterfeit must be so well-made that it can fool collectors is a bit too strong in my opinion. Consider for example the Italian counterfeit of the TMWSTW dress cover LP. It is standardly classified as a counterfeit, despite the fact that the labels are so different from those of the original that no serious collector of Bowie records could ever be fooled by that record. In fact, when I first found that record in the 1980s I didn't regard it as an original for a second, but I was quite happy to have it, nonetheless, because it was the first "dress cover" version of TMWSTW in my collection. So sometimes collectors buy counterfeits to serve as a kind of placeholder in the collection - or simply because it is interesting to have an item that is very similar to a rare original release. So I would like to give the following definition of counterfeits:
A counterfeit is an unofficial record that looks (near-)identical to an officially released record and is attractive to collectors because of its close similarity to the original item.
This covers both the collector-cheating counterfeits and those that are deliberately bought for fun. You might consider this lengthy discussion on how to define counterfeits a bit pedantic, but that is necessary to distinguish counterfeits from pirate releases.

3. On pirate releases
One year after the release of the above-mentioned 2014 counterfeit of the dress cover version of TMWSTW an issue in green vinyl was released. No sane collector would confuse this item with an original release, and to say that it's attractive because of its similarity to that original doesn't make sense, either, since the green vinyl ruins any similarity. This is an example of a record that I call a pirate release. All the music on it is also easily available on official records, and it does not closely resemble any official release. Numerous pirate issues of this kind have turned up over the years - my favourite is a 2013 pirate issue of TMWSTW that pretends to be an Argentine Mercury dress cover LP - no such original record has ever existed!
Note, though, that pirate releases needn't always be copies of records that have been officially released in a different form. There are also unauthorized compilations of easily available material, for instance the 1990 South Korea compilation Best of the Best or the 2018 Italy album Davie Jones... and Other Stories, which is a compilation of officially released early material. So I would like to define a pirate release as follows:
A pirate release is an unofficial record that contains only easily available officially released material, but doesn't look (near-)identical to any officially released record.
This definition doesn't rule out any borderline case, such as the above-mentioned Italian counterfeit (pirate?) of TMWSTW, but I think it is a definition that allows a classification of most of the records in question.
You might wonder why I insist on the fact that the music on a pirate release must be "easily available". This criterion is necessary to distinguish pirate discs from bootlegs.

4. On bootlegs
When thinking of bootlegs most people think of unreleased live recordings and unreleased studio material such as demos and outtakes. In fact, bootlegs with that kind of stuff are the commonest ones. But, as in the case of counterfeits and pirates, things are not as easy as they appear at first sight. Consider, for instance, the 1978 single Memory of a Free Festival. It is a straight copy of a 1970 single and it has a cover that even resembles the covers of some European issues of the original single. It doesn't qualify as a counterfeit, though, since there is too little similarity to any of the original releases. However, if this record was released nowadays, it would clearly have to be classified as a pirate release, since you can easily get these tracks in a record shop (or at least on a second-hand copy of some CD or vinyl record on the internet).
However, back in 1978, the situation was completely different. The original singles had sold poorly in 1970, and had been out of print for eight years. And there had been no re-issues (in fact, the first official re-issue of these tracks came twelve years later). So in 1978 it was practically impossible to find these tracks on any official record! This is why I classify this record as a bootleg. So I would like to give the following definition of a bootleg:
A bootleg is an unofficial record that contains material that is not easily available on any officially released record at the time of release.
I think this definition matches most records that are standardly considered to be bootlegs.

5. Are all unofficial releases illegal?
Many people think that unofficial releases are generally illegal, but this is not true. For a long time they were, and in the 1970s and the 1980s busting bootleggers was one of the primary hobbies of the record industry. However, the reason for the illegal status of most bootlegs is that they contain copyright-protected material, and copyright can expire... This first happened in the late 1980s in the case of the so-called "protection-gap bootlegs". The exact length of copyright-protection of a recording depends on the laws of individual countries, and in some European countries, e.g. Germany, Luxembourg and Italy, the copyright of certain recordings (studio recordings might be treated differently than live recordings, recordings made outside Europe might have a different status than those made in Europe, etc.) expired. This led to the release of several Bowie albums such as Letter to Hermoine and Live USA, which were perfectly legal items in their respective countries of manufacture, and were available at regular record stores. However, due to the free trade inside the European Union they were exported to other countries where copyright law was different and where these albums were, strictly speaking, illegal. To end this confusion the "protection gap" was closed around 1992, and these records were proclaimed "illegal" retrospectively. This strange legal arrangement makes it problematic to re-sell such items nowadays. I've never heard of anyone getting into trouble for selling a Bowie album of that kind, but be careful if you want to sell a Metallica "protection gap bootleg" - you might hear from their lawyers!
For about twenty further years the situation was clear again: unofficial releases were illegal, but from about 2010 onwards the "protection gap" started to open up again. Obviously, the copyright of certain recordings, especially TV and radio broadcasts, had expired again, so that now (2021) you can buy a considerable number of Bowie bootlegs from online shops like Amazon! In fact, some of these are not bootlegs in the above sense, but even pirate releases: for example you can buy a triple album of the Montreal 1987 show that competes with the official release of that show. And it seems to be perfectly legal. Despite this, I sometimes call such items "semi-legal" releases, to point out their strange nature.
Counterfeits are still generally illegal, but, after all, it's just a matter of time until the copyright of regular albums will have expired, too. So in a few decades you will probably be able to buy a home-made Ziggy-album at a regular record shop...

6. Are bootlegs "immoral"?
Legal questions aside, two arguments against bootlegs are often put forward: first, poor quality bootlegs can damage an artist's reputation, and, second, bootlegs take away an artist's control of his own music.
The first argument is mere nonsense. I have never heard of anyone blaming Bowie for the poor sound quality of a bootleg, all complaints have always aimed at the maker of the record.
There is something to the second argument, though. If Bowie had wanted us to hear, say, Tired of My Life, he would have released it, wouldn't he? But then again, it has always been that kind of recording that has kept the "fan spirit" alive during long periods without any new official releases. Sometimes I wonder if bootlegs aren't the best promotion an artist can get... Ultimately, you must decide for yourself if you accept or reject bootlegs and other unofficial releases.

7. A final remark: Anyone can make a bootleg!
Frifelt, who must have an impressive collection of Bowie bootlegs, writes a lot about their different editions. I wouldn't care too much about that. Even using Frifelt's impeccable research on the topic, I haven't been able to determine the exact editions of my two version of the Dollars in Drag album. Bootleggers just take a record and put it in some cover. If they have got records left, but no covers, they create a new cover. And anybody can do that! I have got an album called Göteborg Memories, which is simply one half of the Serious Moonlight in Gothenburg double album. Maybe somebody just accidentally destroyed one of the records and created his own cover for the other...?
To illustrate the point: decades ago I bought a white label album that had "David Bowie" scribbled on one of its labels. Ultimately, it turned out to be half of some version of the Slaughter in the Air double LP. Maybe I should create a nice insert. I could choose a great photo from the time, think up a good title, print out that artwork, make an enlarged copy on a copy machine (computer prints are too easy to identify), put it on a neutral sleeve (maybe even a spare old one from the 1970s)... and here we are: I would have created a mega-rare new Bowie bootleg! I could think up a good story ("Only very few with this insert were made..." - which would even be correct!), and sell the result for 50 Dollars on Ebay.
I won't do that - my "collector's pride" keeps me from doing so. But others could...
The fact that anyone can produce a bootleg has lead to another weird development. Bootlegs have always been available in smaller quantities than official records, and there have always been very limited "special editions". But it has usually been the bootleggers' aim to distribute their products to a wider range of fans. In recent years, however, more and more records pop up (in listings, not on the market, that is) that were made in quantities of only twenty or even ten copies. I can't seriously consider these to be bootleg releases, since you can hardly say that something has been "released" if it has been distributed only to a handful of people. It's a bit as if I wrote a text, made ten copies, distributed these to my friends, and then claimed I had published a very rare book...
But of course it's up to you if you are willing to spend large sums of money for such a home-made item (provided you ever get the chance to find one of these)...

- a hole has been drilled or cut through the cover;
- there is a saw cut through the spine of the cover (this was even done to the jewel cases - and back inlays - of CDs);
- one or more corners of the cover have been cut off;
- a hole has been drilled through the labels of the record (especially in the case of US singles);
- there is a stamp on the cover (German "R" stamps).

None of these things indicate a rare record, a promo, or anything like this. On the contrary, these damages show that the record is an overstock copy that couldn't be sold at the regular list price. Here's the reason:

A newly released record is sold to customers at a relatively high price. Which is no wonder, since record companies want to maximise their profits. This means that retail stores also have to pay a relatively high list price for the records (let's call this "price 1"). And, of course, the companies must provide a certain number of records because otherwise customers would have problems in obtaining a copy of that hot new record. But what happens if the record isn't a success and all these copies can't be sold?
In many countries, including the USA, retailers had (and possibly have - I can't say) the right to return such copies to the wholesaler and, ultimately, the record company, and reclaim the original wholesale price ("price 1"). There were two things that the record companies could do with such records: either they could destroy them, and recycle the vinyl (which often led to poor quality pressings made from that stuff - especially in the 1960s and 1970s), or they could try to sell them again. In the latter case the records were offered to retail shops at a much (!) lower wholesale price ("price 2"), so that they could also be sold at a very low retail price. Such records were sold to retailers without the right to return them to the wholesalers (and, ultimately, the record companies).
So far, so good. But who could prevent a tricky record store owner from returning a "price 2" copy and demanding the original "price 1" wholesale price? The only way to avoid this was to mark the overstock records in some unique way. The ways they found are listed above.
I'm not sure if this is the entire story. In the early 1980s, for example, Italian cut-out copies of Iggy Pop's Soldier album were available in heaps in Germany. I can't imagine that the Italians expected the Iggy album to be the next chart-topper. So I think that second pressings were sometimes made with the explicit intention to sell them as cut-outs. In this way you could make at least some money with a record the sales of which were below those expected by the record company.
Be that as it may, in any case a cut-out does not indicate anything special, but either an overstock copy or - possibly - a cheap second pressing!

It is still unknown when exactly (that is which month) the original German Mercury issue of The Man Who Sold the World was released, but the standard assumption is that record was issued in 1971, like its Australian and UK counterparts. This assumption has been challenged by a strange piece of information I came across on Discogs. Discogs claims that the album was released in 1972, and the reason is given in the description of the record: "Edition with the small pressing ring around the spindle hole, so not pressed and released prior to summer 1972." An official Mercury album by Bowie released in summer 1972? This sounded so absurd to me that I decided to look into this issue a bit more closely, especially because Discogs has become a quite influential website. And I think I can prove that there's no reason to part with the "traditional" view that the album was released in 1971.

Facts

Let's start with a few facts about the record. As pointed out in my entry for the album, there are two different pressings. First, there is an issue with two large centre rings and the following information in the run-out grooves:
- side 1: 10 AA6338041 1Y 320 plus A1 and 1 in different places
- side 2: 10 AA6338041 2Y 320 plus A1 and 1 in different places, plus a faint scratched A4 575 close to the edge of the label
Then there is a pressing with a small centre ring on side 1 and a large ring on side 2. This issue has the following information in the run-out grooves:
- side 1 (small ring): 10 AA6338041 1Y 320 plus a B in a different place
- side 2 (large ring): 10 AA6338041 2Y 320 plus A1 and 1 in two different places

It is not known if the two versions were pressed at the same time (with different machines) or if they were pressed at different times. If the latter is true, then the variant with two large rings probably came first, since white label promo copies have the same label moulding and the same matrix information as these.
However, does the fact that there are copies with small rings mean that they were actually pressed as late as summer 1972?

The legal situation

Let's consider the legal situation. By September 1971 Bowie's contract with Mercury had been terminated, and the deal included that he was given back the rights to his two Philips/Mercury albums. He signed his contract with RCA around the same time. However, it is not exactly clear when the master tapes of the two albums were returned to Bowie/RCA, and some sources claim that Mercury retained the right to keep the master tapes until September 1972. If this is true, then Mercury could have pressed albums in 1972. It is, however, quite unlikely that they did, as the next paragraph will show.

The rarity of Mercury pressings of TMWSTW

But what about the centre rings?

Let's return to the original reason for my starting this discussion - the argument that there can't be a German Mercury album with small centre rings pressed before summer 1972. I will show that this is plainly wrong!
First of all, it is not entirely correct that the album has a "small pressing ring around the spindle hole". As I wrote above there are both copies with two large centre rings and copies with a small ring on one side and a large ring on the other. The latter show that small and large rings were produced at the same time.
But still the claim that small rings can't be found before summer 1972 could be correct. However, Mercury wasn't an independent record company at the time but a daughter label of Philips, who made and distributed their records in Germany. So the claim would only be correct if there wasn't any Philips-made record with small centre rings prior to summer 1972. And this is incorrect.
Not only was Philips a label in its own right, but also a major distributor for other labels in Germany in the early 1970s: Vertigo, Charisma and Island (and possibly more) all had their records made and distributed by Philips. This can easily be seen from the Philips-style catalogue numbers: a four digit block always starting with "6" followed by a three digit block, which is in turn followed by a letter (the price code) on the cover (although there are a few exceptions to this pattern).
So to prove the above claim wrong, we must find a Philips-made record that (a) was definitely pressed in 1971 or earlier, and (b) has a small centre ring. And, oddly enough, this is were old Bowie pals Mott the Hoople come into play.
Mott the Hoople's third album Wildlife was originally released on the pink "i" Island label in Germany. The catalogue number was 6339 031 D, so the original pressing was definitely made by Philips. In the course of 1971 (June, according to Discogs) Island changed their German distributor from Philips to Ariola, and the record was quickly re-released on the "pink rim" Island label with a new, Ariola-style catalogue number (86 696 IT). Note that all re-issues and re-pressings of early Island records in Germany were also provided with new Ariola-style numbers. So regardless of wether the change of distributor was actually in June, we can safely assume that all Philips-made copies of Wildlife are pressings from 1971. And the copy of the album shown on Discogs has small centre rings. With respect to the present topic my personal copy is even more interesting: it has a small ring on one side, and a large ring on the other - just like The Man Who Sold the World. So Philips did make records with small centre rings in 1971!

Conclusion

To sum up, then, there are no convincing reasons for 1972 as the year of release of the original German The Man Who Sold the World. On the contrary, there are good reasons to assume that all Mercury pressings of the album, including the German one, were made before 1972.