(Note: this article is long and complex, and, in the end, not really crucial to our understanding of the Palaeologan coinage - so, unless you are a glutton for punishment, you only need to read the first two  sections below and the conclusion - R.G.).

The hoard of 2000.

In Numismatic Chronicle 2001 Simon Bendall published a reconstruction of a find of Palaeologan bronze coins which had presumably been discovered the previous year, i.e, in 2000, and which had been sold through a number of dealers in Europe and also H.J. Berk's in the U.S. The find consisted of about 380 coins, most of them issues of Thessalonica, primarily the later trachea of Andronicus II, together with some types normally attributed to Andronicus III. 

This find was very interesting in itself, as reported hoards of coins (particularly bronze coins) from the Palaeologan period are rare, but the real importance of Bendall's find becomes evident when we compare it with Longuet's well known "Thessalonika" hoard. At first sight, both Bendall's and Longuet's hoards would seem to be close together in time - Longuet's hoard is usually assumed to date from c.1342, while Bendall's seems, at first sight, to date from the middle 1330's (see below). We therefore might expect the two finds to be very similar in content, but when we compare them we see immediately that they are in in fact very different - Bendall's hoard is mainly Andronicus II's, with a few Andronicus III types, whereas Longuet's is mainly Andronicus III's, with only a handful of Andronicus II's. Apparently, then, the two hoards must actually cover rather different, although overlapping, periods of time.

This provides us with both an opportunity and a problem. As explained in another article on this site ("The Thessalonican Trachea of Andronicus III"), the differences between the hoards can in fact be very useful, since it enables us to determine which of the later Palaeologan types types to assign to Andronicus III, and which to Andronicus II. On the other hand, given that the two hoards seem prima facie to be fairly close together in time, how do we explain these differences in the first place? At first sight, this might seem a relatively minor problem (and one which was apparently not considered, explicitly at least, by Bendall), but it will, as we shall see, take us on a long journey, involving many different and difficult questions.

Actually, as we will also see, another important question will need to be considered as well, namely, the nature of the find. At first sight, it might seem that we are dealing with a simple discrete "snapshot" hoard here, i.e, a hoard formed and deposited at a particular time, so that it reflects the make up of the coinage circulating at that time. In fact, we shall see that there are significant problems with this assumption, and we will ultimately have to consider the possibility that the find is actually more an accumulation than a discrete hoard..

But to begin with, let us treat the find as a hoard, as Bendall seems to have done, and try to work out its implications on that basis.

The date of the 2000 Hoard.

Whether or not Bendall's find is a simple hoard, at first sight it would appear that it must date from the mid 1330's, since it includes some, but not all, of the commoner types which are usually attributed to Andronicus III (that is, the types found in the "Thessalonika" hoard, which contained virtually all of the nominally Andronicus III issues). But oddly, Bendall prefers to date the hoard to the later 1320's, and assumes that the Andronicus III types are intruders, or not actually issues of that ruler. Why? - well, this is not really made clear. Hoards are normally deposited, or at least abandoned, during periods of unrest, and so, if Bendall wants to assign an earlier date to his hoard, then a number a possibilities are available to him. For example, one possibility is that the find was deposited in the early 1320's, during the various disputes between the Byzantines and the Bulgarians, but Bendall doesn't opt for this period. Another, and perhaps more likely, period would have been the wars of the early 1330's between Byzantium and the Serbs and Bulgarians, but Bendall doesn't opt for this either, and in fact he suggests that the hoard dates from the end of the joint reign of Andronicus II and Andronicus III, or the earliest years of Andronicus III (Num. Chron. p. 273).

Now Bendall gives no reasons for this odd late 1320's dating, but perhaps it was based on an assumption that the common assarion S.2481 was absent from the hoard (or nearly absent - in fact, as noted below, one or two examples of this type seem to have been present). However, whether or not this was the case, I can suggest another argument for a dating somewhere in the 1320's, deriving from the differences between the hoards mentioned above.

The 2000 hoard versus the Salonika hoard.

As we saw at the beginning, Bendall's and Longuet's hoards have very different compositions, and we now need to try to account for this.

Prima facie, if we assume that we are dealing here with normal times (i.e, with new types being issued at a reasonably constant rate, on a reasonably regular basis), and normal hoards, such as a merchant's hoard, formed and deposited at a particular time, then the almost complete difference between the two hoards would seem to mean that the 2000 hoard has to predate the Salonika hoard by some time - 20 to 25 years at least. Then, since Bendall has always assumed (e.g, in LPC) that the Salonika hoard dates from c.1342, it would therefore seem, on the stated assumptions, that the 2000 hoard would have to date from c.1320 at the latest. But given the Andronicus III types in the hoard this would seem unlikely, and hence, if we are to keep Bendall's hoard in the 1330's, or even the later 1320's, we must seek other explanations for the difference between the hoards.

Basically, there are two ways out of our dilemma - either one or other of the hoards is not a normal hoard (in the sense just defined), or the date of one or both of them is wrong.

Let's start with Longuet's hoard - as we have seen, as compared with Bendall's hoard, Longuet's  noticeably lacks Andronicus II types. Is this deficit real, or has the hoard been modified in some way, so as to reduce the number of Andronicus II types, so that it no longer reflects the composition of the currency at the date of its deposition? We can envisage a number of scenarios, both ancient and modern, under which such a modification might have occurred.

To start with the obvious, maybe the Longuet hoard is the residue of a more general find that had been picked over for the known Andronicus II types before the rest was offered to Longuet by the dealer in Thessalonica. This is certainly possible, although surely anyone knowledgeable enough to have done this in the 1930's would have also recognised that the new types were even more important (i.e, valuable) than the Andronicus II's. In any case, we are told that the find was brought in to the dealer by a peasant, who would hardly have such knowledge, and why would the dealer conceal the (hypothetical) Andronicus II types from Longuet, a well-known collector of such types?

So maybe the answer is much older - perhaps the Byzantine administration was forced, for some reason (such as a recoinage), to withdraw most of the existing bronze coinage from circulation some time in the 1320's. However, there is no record of such a proceeding, and in any case it's difficult to see when it could have occurred, as it would obviously have to post date the 2000 hoard, which, prima facie at least, extends into the 1330's, overlapping the Longuet hoard. This idea therefore would have to combined with Bendall's theory that the Andronicus III types in the 2000 hoard are intruders, adding one ad hoc assumption to another - not impossible perhaps, but overall not very convincing.

Perhaps, then, the explanation is spatial rather than temporal - perhaps the hoard actually came from northern Thessaly, rather than Macedonia. In the later 13th and early 14th century this area, known to the Franks as the "Duchy of Neopatras", was under the control of Epirote rulers. In 1319 it fell to the Catalans of Athens, who (presumably) would have introduced Frankish coinage, but was eventually retaken by the Byzantines in 1333. After this takeover, the Frankish coinage would have been replaced, over a period, by (mainly) the post 1333 coinage of Andronicus III, and this could well have yielded a mix of types something like that actually found in Longuet's hoard. This idea is perhaps supported by the presence of several Epirote deniers in the hoard, but overall the scenario seems unlikely. (Nonetheless, although this idea might seem far fetched, it is not totally out of the question. It might well account for the fact that quite a number of coins in the Longuet hoard were struck on Epirote deniers - perhaps these overstruck types were mainly sent to Thessaly as an area familiar with deniers).

So perhaps it's not the nature of Longuet's hoard which is the problem, but its date. It is usually assumed that this hoard dates from the establishment of the zealot commonwealth in Thessalonica, i.e, 1342, but if we assume that the date is closer to 1350 than 1340, then we certainly begin to explain the lack of Andronicus II types (particularly if we also assume, as we will do below, that the commoner Andronicus II types predate 1320). Now it is possible, and in fact quite likely, that the hoard dates from as late as 1345, when the Serbs overran most of Macedonia, but there seems to be little to justify a date any later than this, and in any case if the hoard dated from much later than the mid 1340's, so that it reflected the coinage circulating around, say, 1350, then from its composition we might have expected to still see a reasonable number of Andronicus III types in the Pella and Serres hoards of the 1360's, whereas in fact there are almost none. Would a date of 1345 rather than 1342 solve our problem? - well, it would help a little, but hardly enough, as it is still not even a decade after the apparent date of Bendall's hoard.

In any case, it is not impossible that this hoard predates 1342 by several years, since, as explained elsewhere (see "The Thessalonican trachea of Andronicus III") , it does not include certain reasonably common types such as S.2525 which could well be issues of Andronicus III. This would of course make our problem even worse.

Overall, then, the idea that the problem lies with the Longuet hoard doesn't seem to be very promising, so perhaps we need to consider the 2000 hoard.

In effect, we have been assuming so far that coin production continued more or less steadily not just throughout Andronicus II's reign, but well into the 1320's. However, it is clear that this did not occur - we know (from the relative scarcity of most Andronicus III types) that coin production in Thessalonica dropped drastically some time before the accession of Andronicus III, and was only partly revived in the 1330's. In fact, it is generally assumed that production dropped off some time around 1320, perhaps due to the disputes of the period between Thessalonica and Constantinople, or the constant local wars with Bulgaria and Serbia, so that all the commoner types of Andronicus II date from the earlier to middle, or late middle, periods of his reign. This then, it might appear, is the real reason for the lack of these types in the Longuet hoard.

In effect, since under this scenario the bulk of the coins and types in Bendall's hoard derive from before 1320, then that hoard would effectively date from that time, at least as far as comparison with Longuet's hoard is concerned, and this, as noted above, is just what the comparative hoard statistics require. (In fact, it's worth noting here that Bendall generally assumes that Andronicus II issued no coins at all at Thessalonica in his second sole reign (meaning in this case 1320-25), and while this is not impossible, there seems to be no specific need to assume a complete cessation of issues in this period - in fact the large number of different types in Longuet's hoard suggests that some of them could well be issues of the mid or even early 1320's. In reality of course these would have been issues of Andronicus III, since Andronicus II no longer really controlled Thessalonica).

This idea sounds very promising, but while it might explain the content of Longuet's hoard, by itself it still doesn't provide a complete solution to our problem with Bendall's find. If this find is only seven to ten or so years earlier than Longuet's (as would seem likely at first sight), then we would still expect its composition to be not all that different from Longuet's (no matter how the production had varied over time), whereas of course it is in fact almost totally different, with large numbers of Andronicus II coins drawn from many different types. To put it another way, if all of the the common types from before 1320 were still circulating in bulk in the early to middle1330's, when Bendall's hoard seems to have been deposited, then there should still be plenty of them still circulating in the early to mid 1340's, the probable date of Longuet's hoard.

From yet another perspective, the plethora of types in Bendall's hoard is a problem in itself - there are at least 34 single emperor and 7 two emperor types in the hoard (in fact all of the commoner Andronicus II types are present), so if this hoard dates from the 1330's, but most of its coins date from at least 15 years earlier, it is difficult to see how so many examples of so many different types could have stayed in circulation for so long.

So perhaps Bendall is right - maybe the 2000 hoard dates from the mid 1320's, rather than the 1330's. Combined with the assumption of a drop off in production around 1320, this might (just possibly) explain the difference between the two hoards, although in reality I think we still have the same basic problem as before - i.e, if a hoard deposited in the later 1320's was full of Andronicus II types, would one deposited only 15 or so years later contain virtually none? Could the composition of the circulating coinage have changed almost completely in such a relatively short time? It would seem unlikely, and now of course we have another problem - how to explain the apparent inclusion in the 2000 find of the Andronicus III types like S.2486 and 2487.

Hoard versus accumulation.

This leads us to reconsider one of the key assumptions made above, namely that the two hoards were essentially formed and laid down at a particular moment in time, and with Bendall's find at least there would seem to be a strong argument that this may not have been the case.

It is instructive to compare the 2000 find with the Thessalonican coins from the reigns of Andronicus II and Andronicus III excavated by Dochev at Turnovo, which roughly match the 2000 hoard in overall numbers, and which presumably represent a reasonably random sampling of the issues of Thessalonica over the period of interest.

Overall, the distribution of types in the 2000 find closely matches that of the coins found at Turnovo. (The actual figures can be found in the article "Sequencing the Thessalonican Trachea" on this site). The main differences are that the 2000 find includes only a few pre-1282 issues, and seems deficient in types from the earliest years of Andronicus II's reign (although these were not common at Turnovo either), while at the upper end of the time frame the 2000 hoard (at least as reported) lacks the late Andronicus III types S.2484, 2497 and 2501, all of which were found at Turnovo. Overall, all this is consistent with Bendall's hoard being the remains of an accumulation of some sort, rather than a discrete "snapshot" hoard.

If so, it does not necessarily mean that the accumulation was the result of stray losses - it could perhaps have been the savings of a household put together over a period beginning some time after c.1300, and terminating in the mid 1330's (or perhaps even later*), when the coins were presumably finally abandoned. In other words, this could have been a domestic rather than a commercial hoard.

Bendall justified his assumption that the find was a hoard (rather than a random accumulation of stray losses) by appealing to the similar patination of the material, and claiming that this indicates a hoard, but a "quasi-accumulation" as defined above would also show this. On the other hand, it seems fair to say that there is some evidence of a progressive increase in wear between the earlier and later types, which at first sight would seem to support a discrete "snapshot" hoard hypothesis. For example, the earlier Andronicus II & Michael IX types all show medium to heavy wear, which might be taken to indicate a discrete hoard, but in fact it's also be consistent with an accumulation put together over the period indicated above.

The accumulation idea allows us to explain the numerous different types in the 2000 hoard by spreading their accumulation over a relatively long period. This then allows us to adopt the assumption that the commoner Andronicus II types date from the middle of Andronicus II's reign without the difficulties previously raised, which in turn allows us explain the lack of Andronicus II's in the Longuet hoard. Overall, then, these two ideas allow us to resolve the Longuet versus Bendall problem in a simple manner, without any drastic assumptions about the nature of Longuet's hoard.

However, we are perhaps not totally out of the woods yet - we note that the composition of the 2000 hoard is basically much the same as the "2004" find reported in another place (see "Some Recent Palaeologan Finds"), another recent find (or more likely a number of finds) from the Balkans, so are we too assume that all of these various finds are also domestic rather than commercial hoards? It's not impossible, but the compositions of the various finds seem to be much too similar to each other, and also to Dochev's figures, for them all to be the product of domestic savings by different households over a long period (since it's unlikely that their savings patterns would be the same). Perhaps then there is still some other explanation of the composition and formation of these finds yet to be found.

*  It is possible that the termination date is even later - S.2497 is not noted as being included in Bendall's hoard, but there were in fact two examples of the type in Berk's sales in 2003-4 (BBS134-597 and BBS136-365) which could conceivably have been part of the hoard. At first sight this might seem unlikely, since by my calculations most of Berk's portion of the hoard was sold by BBS128, and Berk's reported no 2497's in their part of the hoard (and they're hard to miss), but I think it's a real possibility. If these coins were in fact part of Bendall's hoard then they (and the hoard) could date from as late as c.1240. This conclusion would be further reinforced if we include BBS128-719, an example of Longuet 2, in the find. (Bendall would of course have to make all these coins intruders, along with all the rest of the late types).

The S.2486's.

We now need to consider some further problems with Bendall's dating of the 2000 hoard to the 1320's.

Why did we originally date the 2000 hoard to the 1330's? The key point that we have to understand is the dating of Longuet 14, i.e, S.2486. This type is normally assigned to the 1330's because some of the S.2486's in the Longuet hoard (and also some of the 2487's and, supposedly, one of the 2483's) are overstruck on debased versions of the billon deniers of John II Orsini of Epirus. Now on these coins John is styled as Despot, a title he apparently did not receive until 1328-9, so that Longuet 14 must date from 1329-30 at the earliest. Also it is reasonable (says Bendall) to assume that the deniers used by Andronicus III were acquired during his various wars in the early 1330's (since we know that Orsini style deniers are notably found in Bulgaria), which would put the S.2486's, or at least the ones in the Longuet hoard, well into the 1330's.

(The source of the debased Orsini deniers used in Andronicus III's coinage is a bit of a mystery - Orsini deniers, which are billon coins of varying alloy quality, are scarce types today, but they have turned up at several sites in Bulgaria. Metcalf has suggested that these finds are trader's caches, but overall the deniers seem to have been rather too common for that. In fact, judging from Dochev's excavations at Turnovo, they appear to have been one of the commonest small coins of their time in that city at least, where many were found as strays, in both finer and debased form. Hence perhaps Bulgarian sources could in fact have supplied the Byzantines with the large number of deniers that they evidently acquired. Another possible source of these coins might be Neopatras, which as we have seen, was reoccupied by Andronicus III in 1333 - it is possible that Epirote deniers, or debased copies of them, were used there, although there is little evidence as to the make up of the coinage in that area at that time. On this latter hypothesis, the S.2483's, 2486's and 2487's would then date from the mid 1330's).   

Anyway, if Bendall wants to assign the 2000 hoard to the 1320's, he has to account for the various Andronicus III types in the hoard, particularly types like S.2486 and 2487, which presumably date from the middle of Andronicus III's reign.

For the S.2486's Bendall invokes two interesting arguments. Firstly (according to him), all the examples of this type in the 2000 hoard are significantly different from those in the Longuet hoard, and hence can be dated earlier. Specifically, he argues that the 2000 hoard examples are more scyphate than their Longuet counterparts, and also that none of them are struck on Orsini deniers. Therefore (Bendall concludes) there must have been two issues of this type, one by Andronicus III, and an earlier one by Andronicus II. Now there is kernel of truth in this idea, but the situation would seem to be rather more complicated than Bendall imagines.

To begin with, many of the so-called "S.2486's" in the 2000 hoard are indeed examples of an earlier issue, but not of S.2486. Rather, as explained in the Note "S.2315 of Michael VIII and S.2486 of Andronicus III", they are actually examples of the so-called "starred" variant of S.2486, a type which is actually nothing more or less than S.2315, a coin assigned to Michael VIII in LPC and Sear, but which is undoubtedly an issue of Andronicus II, and clearly quite separate from the real S.2486 (i.e, Longuet 14) of Andronicus III. Thus to this extent Bendall was right - there actually are at least two separate versions of  "S.2486", although they are not two issues of the same type, but separate issues of two similar but distinct types.

But the story doesn't end here - even after removing the obvious S.2315's from consideration, we still may have two versions of the real (i.e, non-starred) S.2486 in the 2000 hoard.

Firstly, we note that most examples of this type are struck with flat dies on flattish flans, as in the Longuet hoard. However, sometimes the flans at least are more scyphate, as is typical with the later Andronicus II types (for a striking example of S.2486 struck with a flat die on a scyphate flan see Lot 301 in the "Despot" sale of 2006 - LHS Numismatics Sale 137). Are these scyphate flan types earlier issues? - it's not impossible, but it seems more likely that all the flat die types, whatever flans were used, date from the early 1330's, and form a single transition type between the issues of the 1320's and the lighter and flatter issues of the later 1330's. It thus seems reasonable to assume that these varying issues date from the period when the Orsini deniers were first being used in quantity as flans for the Byzantine issues.

However, as opposed to flat dies on scyphate flans, we also sometimes see examples of non-starred S.2486's struck with clearly scyphate dies (on scyphate flans). Where to assign these particular types is unclear - there was at least one example in the 2000 hoard (BBS117-515) and also several more in the hoard of 2004 mentioned earlier, another find (or finds) of mainly Andronicus II types which does not appear to contain any real (i.e, flat die) S.2486's. So perhaps there really was an early scyphate version of S.2486 after all, as Bendall suggests (although if so it was possibly simply S.2315 without the stars). On the other hand these types seem to be fairly light, so that they may still be issues of Andronicus III after all.

Returning to Bendall's hoard, if we ignore the (obvious) starred types, are we left with any possible real S.2486's? By my count there are at most eight possible candidates remaining, namely coins 130 and 131 in the Num. Chron. article, two coins in the Gorny & Mosch part of the hoard (Sale 104, lots 1490-91), and the Berk's offerings BBS 117-515, 117-521, plus (perhaps) 128-718 & 134-569 - if indeed the latter two are part of the find. (Note also that I don't include BBS134-593 here - this coin was offered later on the Rudnik site, and from its appearance is likely not part of the hoard). Of these eight coins, two (BBS117-515 and 134-569) are indeed more or less scyphate, and hence may possibly be earlier issues, and GM104-1490 is struck on scyphate flan, but the rest are quite as flat as the Longuet examples (i.e, they are struck with flat or flattish dies on more or less flat flans).

Now, because of small flans or damage we can't really be sure that all these flat coins really lack stars, and in fact, of the five flat coins, Bendall's No.131 and GM104-1491 could still possibly be starred types. But even if we leave these two coins out, we still appear to have two (or three including BBS128-718) flat non-starred types, i.e, coins which are prima facie regular issues of Andronicus III, plus GM104-1490, a flat die on scyphate flan type, and hence probably also an issue of Andronicus III. Thus, contrary to what Bendall wants to assert, it would appear that at least some of the non-starred coins could well be real S.2486's, which is of course a serious problem for an early dating of the hoard, no matter how it was formed.

(Note that Bendall suggests that the Gorny & Mosch coins are too heavy (at 1.61 and 1.63 grams) to be issues of Andronicus III, but  in fact their coins are no heavier than several of the S.2484's, 2487's and 2497's in the Longuet hoard - also the Despot S.2486 mentioned above weighs 1.77g, while Gr.1469 (an S.2486) weighs 2.00g, just to mention two examples. The "forked" beards that Bendall notes on two these coins are not all that significant - similar beards are sometimes found on other Andronicus III types - e.g. the S.2487 examples BBS134-594 & 147-374).

To sum up this laborious analysis, I think we can say that while a case for a separate issue of scyphate (unstarred) S.2486's under Andronicus II can be made (perhaps as variants of S.2315), we still may have some regular S.2486's in the 2000 find to account for.

Other "Andronicus III" types - S.2487's and S.2483's.

In fact, the S.2486's are just the beginning of Bendall's problems. For example, the 2000 find also reportedly included quite a number of (nominally) Andronicus III types, namely S.2482, 2485, 2488, 2490, 2492, 2496 and Gr.1473, which would all seem to be a difficulty for Bendall's early date. However, with these types the problem isn't as bad as it seems. For various reasons which are explained elsewhere (see "The Thessalonican trachea of Andronicus III") it's reasonable to assume that neither S. 2485 (meaning here the standing emperor type Longuet 15) nor S.2488 were in the hoard at all, and that the others in the list above are either Andronicus II's, or early Andronicus III's at the latest. Thus, for these types at least, we can accommodate Bendall and squeeze the hoard date down to the late 1320's, or even earlier if we are prepared to reassign all the relevant types to Andronicus II or the joint reign of the two Andronici.

More seriously for Bendall though, the 2000 hoard also includes several S.2487's (five in fact, according the table on p.275 of Bendall's report*). Now all these coins are apparently quite flat (or nearly so, like virtually all examples of this type), and hence they can't be from any earlier issue, and so, since Bendall wants to date the 2000 hoard to the 1320's, he has to resort to an old standby and call these coins intruders. He justifies this by asserting that the 2487's have a rather different appearance to most of the other coins in the hoard (although it seems he had not seen all of the Berk's examples at the time of his article). Now I can't comment on the UK coin that Bendall did see, and he may well be right about it, but while one or two of the others certainly show some extra corrosion, they don't all look like obvious intruders to me. In particular, the Gorny & Mosch example (GM104-1492) apparently has the same "green" patina as many of the other hoard coins. (It's perhaps worth noting here that in my experience S.2487 has often seemed an odd type - for example, the style is often rather crude, and the reverse is often upright rather than inverted). Overall, I see no convincing reason for not taking  the 2487's, or at least some them, to be part of the find.

(* But note that some of these coins may not actually be from the hoard - the table on p.275 lists five S.2487's in the hoard, made up apparently of one coin from the UK, three from Berk's, and the Gorny & Mosch coin. The UK coin is specifically rejected by Bendall as an intruder, and is not described in the body of the report. Berk's offered one 2487 for sale in 2000 as BBS117-517, and two more some time later in 2003, as BBS134-594 & 595, but whether the last two coins were part of the original three is not clear. On the other hand, the Gorny & Mosch coin certainly appears to be part of the hoard, as noted earlier).

Again, Bendall assumes that there were no S.2483's in the hoard, but it appears that in fact there may have been one or two of these types present. According to Berk's (as quoted by Bendall) there were no S.2483's in their part of the hoard. However, since Berk's started offering the hoard material (in BBS116) three examples of S.2483 have in fact been offered by them (as BBS120-543, BBS128-716 and BBS134-592). So were some or all of these S.2483's part of the original hoard? By my count most of the hoard coins in Berk's possession seem to have been sold (or at least offered) by BBS128, but some apparently had not, even allowing for errors in the Berk's list, so it seems quite possible that the 2000 hoard did in fact include at least one S.2483, which again might seem to imply a 1330's date for the hoard (but more about this type later). Also, the S.2483 in the CNG 53 sale of March 2000 {part Lot 1948} may possibly have been from the 2000 hoard - it has the dark grey green patina typical of the hoard, and so, apparently, have three or four other Michael VIII and Andronicus II types from that sale.

Finally, it should be noted that Bendall also needs to invoke the intrusion theory for at least two other presumed late types which apparently showed up in his hoard, notably the Longuet 2 type (BBS128-719?) and the rare DOC V 1190 type (BBS117-524), which is actually the same type as Longuet 22. (The DO 1190 doesn't actually seem to get a mention in the hoard report as printed, but was presumably part of the find). However, we can't put too much store on these types, since their current dating  (to John V) is conjectural, and the dating of DO 1190 at least is almost certainly earlier. In fact, the fact that they both appear in the 2000 hoard (and of course in Longuet's hoard) implies that both these types are issues of Andronicus III at the latest, or perhaps the earliest years of John V - see "The Thessalonican trachea of Andronicus III". (Note also that Bendall makes the S.2465's in the hoard intruders, based, I can only presume, on their appearance - chronologically there is no real need to assume that these types were not part of the hoard).

In my opinion, all this really means one thing - if we are to take the 2000 find to be a hoard, or even a quasi-accumulation, and then date it to the 1320's, then we have to assume that all the later Andronicus III types in the find are intrusive, i.e, not just the 2487's, but also the flat die 2486's, and any 2483's, 2497's and Longuet 2's which may be part of the find. Now this is not completely out of the question - after all, we are dealing with a "reconstructed" hoard here, not a properly documented original find, and the hoard as we now have it could well include at least some coins from other finds, as in fact appears to be the case in respect of the very similar "2004" hoards of mainly Andronicus II types. But while there may be genuine reasons for excluding some of the S.2486's and 2487's from the present hoard, it does appear likely that at least some of these types are still present in the hoard.

Basically, therefore, I think that for the present we have to take the 2000 hoard at face value and work from there. Hence if we assume that not all the Andronicus III types can be excluded, then we also have to assume that the hoard dates from when it appears to, i.e, the mid 1330's, or quite possibly even later.
 
The Denier Undertypes.

Unfortunately, we are still not finished with the S.2486's. There is still  the second leg of Bendall's argument for an earlier issue of this type, which turns on the denier undertypes. It is noticeable that while 36% (four out of eleven) of the S.2486's in the Longuet hoard are overstruck on John Orsini deniers, none of the (at most) eight non-starred examples in the 2000 hoard seem to be. (Note that none of the starred examples are either, in this find or anywhere else, since they predate John Orsini by some time).

At first sight this observation certainly seems to support the idea that the Bendall hoard 2486's are drawn from a different (and presumably earlier) population than the Longuet examples, and this argument is supported by the statistics for S.2486's from other sources - of the sixteen other (non scyphate) examples of this type known to me, four, or more likely, five, are overstruck on deniers, a similar percentage to Longuet.

However, while the lack of denier undertypes for the S.2486's in the 2000 find, as compared with Longuet's hoard, is certainly noticeable, the statistics don't necessarily imply that we are dealing with two different issues of this type, particularly given the fact, as detailed earlier, that of the 8 coins mentioned above only 5 are flat die types, and of these only two or three can be confidently described as real Longuet 14's.

And of course, this argument can be applied to the S.2487's as well - since 25% (two out of eight) of the Longuet 2487's are overstruck on deniers, while none of the five 2487's listed by Bendall are. So do we now have to surmise that there were two issues of this type as well? - this is stretching credibility too far (particularly as all the 2487's involved are flat types and hence clearly date from the 1330's).

Once again we only have a few 2487's in the hoard, so the fact that none have denier undertypes is not all that significant. (I note that of eighteen S.2487's from other sources known to me, two are overstruck on deniers (of some sort), proportionately somewhat less than in Longuet's hoard).

Overall, then, assuming that we only have one population of real S.2486's, we note that of the 30 confirmed examples considered here, including those in Longuet's and Bendall's hoards, 9 (30%) were struck on deniers, while for the S.2487's the figure was 4 out of 31 (13%). Given the fact that S.2487 is known overstruck on 2486, this suggests that the deniers were first used as a feed stock for the minting of S.2486, and then to a lesser extent for 2487.

To complete the picture, I note that one of the three S.2483's in the Longuet hoard was struck on a denier. On the other hand, none of the three BBS examples mentioned above were, nor was the CNG 53 coin - but again, the lack of undertypes here is not statistically significant.

In fact the Longuet S.2483 with the denier undertype raises an interesting question. Bendall states (or perhaps simply assumes) that the undertype was an Orsini denier, which could be hard to reconcile with what said above, since on various grounds (e.g, size and fabric) it seems that S.2483 likely precedes 2486 and 2487. However, I have my doubts about the undertype here - of 14 other 2483's known to me, three were indeed struck on base alloy deniers, but it is not clear that the undertypes were issues of John Orsini - in fact two of them showed legends which apparently include "dot A", and hence were possibly Catalan period "Dvx Atenes" types from Athens, acquired, perhaps, from Neopatras at some time. But if the undertypes of the S.2483's are Catalan (rather than Orsini) types then S.2483 could predate 2486 and 2487 by some time. In this case of course the presence or otherwise of this type in Bendall's hoard would be irrelevant to the question of the date of the hoard. 

S.2481 of Andronicus III.

Finally, there is one more type in the 2000 hoard that needs to be considered. The hoard included a dozen or so coins from Constantinople, mainly assaria of Andronicus II and Michael IX, plus a few trachea of Andronicus II. Importantly, there was also at least one example of S.2481, an assarion usually assigned to Andronicus III alone (this was BBS116-615 - there was another example of the type in BBS125 and also others in some later sales, although these may not have been part of the hoard). Bendall wasn't aware of the Berk coin when he initially wrote his hoard report, and probably assumed this type wasn't present in the find, which may be why he puts a date of not much later than 1328 on the find. In fact, provided we accept the assignment of S.2481 to Andronicus III, then the presence of the type puts a lower limit of at least 1328 for the date of the hoard, a date not far below the early to middle 1330's estimate that we started with (unless of course we take this coin to be an intruder as well, or reassign the type to the earlier 1320's - there is after all no conclusive evidence that it really is an issue of Andronicus III, although on the whole this seems a reasonable assumption). 

Conclusion.

Overall, we see that, firstly, the 2000 hoard seems to date from the 1330's, and that there are difficulties dating the hoard to the 1320's, as Bendall wants to do, although this is perhaps not totally out of the question.

Secondly, if we assume that Bendall's find is a simple discrete "snapshot" hoard, then there are serious problems in reconciling it with Longuet's hoard, whether we date it from the 1330's or the 1320's. We are therefore led to the possibility that the 2000 find is more likely an accumulation, possibly domestic savings put together over the first 30 years or so of the 14th century, than a discrete hoard, but there are also difficulties with this idea, so perhaps there is some other solution, yet to be found, to the problem of the disparity between the two finds.

For the moment therefore I feel that we can't definitively determine whether Bendall's find is a discrete hoard or a quasi-accumulation (although I incline to the latter), and I will therefore keep both alternatives in mind when assessing problems such as the dating the later Palaeologan issues (see the article "The Thessalonican trachea of Andronicus III").

Fortunately, however, the question of whether Bendall's find is a discrete hoard or not (and whether it dates from the 1320's or the 1330's) is not in itself crucial to most of these dating problems. For these purposes, all that matters, as is shown in the article on the Thessalonican trachea, is that Bendall's find and Longuet's hoard cover essentially different periods of time, and hence give us samplings of the coins circulating in two different, although overlapping, periods - the exact date of Bendall's find is not really important, and nor does it matter too much whether the find includes S.2486's and 2487's or not, or for that matter any other possible Andronicus III types.

Ross Glanfield

August 2004

Last revised:   8 Apr. '05  (Bendall's hoard as accumulation).
                     10 July '05  (Denier undertypes section revised).
                     26 July '05  (Discussion substantially revised).
                      9 Aug. '05  (Nature of hoard revised again, and discussion rearranged).
                     16 Oct. '05 (Discussion of hoard dating revised).
                      9 Jan. '06   (Flat v. scyphate S.2486's argument revised).
                      9 May '06  (Section on S.2481 added).
                      3 June '06  (Details of S.2486 & 2487 discussion revised).
                     14 July '06  (S.2486 discussion rewritten again).
                     13 Dec. '06 (Nature of hoard and other aspects of discussion revised and clarified).
                     11 Jan. '07  (Section on Denier undertypes revised).
                     23 Nov. '07 (Rewritten with Bendall's hoard as "quasi-accumulation"). 
                     11 June '08  (Possible dating of Bendall's hoard to c.1340 considered). 
                       9 Nov. '08  (Section on Denier undertypes revised again).                            

(The statistics in "The Denier Undertype" section are also kept up to date as required).