The problem.

The bronze issues of John Comnenus-Ducas have always been an oddity amongst the coins of the Latin period, posing a number of problems for numismatists. To begin with, these types come in a wide variety of sizes, ranging from the larger module "Series I" and "Series II" types to the small and very small "Series III" types. But the most obvious difficulty is the fact that there are more than a dozen Series I and II types altogether, and more than two dozen different Series III issues, to fit into the 5-6 years of John's reign (or at most 9-10 years, if we include the periods of John and Demetrius C-D as depots).

Hendy's theory on the three Series.

At present the standard theory, as first proposed by Michael Hendy in his book "Coinage and Money in the Byzantine Empire, 1081-1261" (1969), is that the two larger module series are both official issues of Thessalonica, and that they were issued in parallel, i.e, the Series I types were struck, on essentially a one issue per year basis, through the reign of John C-D as emperor, while the somewhat smaller Series II types were issued in parallel with the Series I types over the same period. It is noticeable (pace Bendall - see Appendix) that these two series don't duplicate each other - i.e, the designs of the Series II types are different from those of Series I (although in some cases we occasionally see smallish versions of the Series I's, between Series II and III in size).

As to why there were two parallel series, Hendy puts forward the ingenious idea that the mint issued large module coins (in return for bullion or old coins) to the administration on one weight scale, and to the public on a somewhat lower scale. However, there is no separate evidence to support this theory, and why the supposed practice manifests itself only in John's reign is not explained*. There seems to be little point in Hendy's scheme anyway - after all, why not just issue one series of coins at different exchange rates? And how were the two only slightly different series supposed to be exchanged in the market? The more I think about it the less convincing Hendy's idea appears, and hence I will consider an alternative approach to the problem below.

The small Series III types divide into into two categories, the "imitative" types, which copy the designs of both the Series I and Series II types, and the "novel" types, for which there are, at least at the present time, no known larger module prototypes.

Hendy's theory on the Series III issues, as now expounded in DOC IV, is that these types (and for that matter virtually all of the small module versions of the coins of the Greek empire of Thessalonica, including those of both Theodore and Manuel C-D), were issued not by the Greeks in Thessalonica, but by the Venetians (somewhere), for use in trade (in Bulgaria presumably). He proposes a complicated scenario, according to which, if I understand him properly, the dozen or so "imitative" Series III types were issued, like their larger prototypes, during John C-D's reign as emperor, while the remaining, and supposedly cruder and/or more imaginative, "novel" types were issued after the demotion of John to despot in 1242, or his death in 1244, and also that the series extended well into the 1250's. (Hendy doesn't say where these types were produced - clearly he wants to say in Thessalonica, but realises that this is unlikely, especially after 1246, and obviously can't come up with a better idea).

*  Actually, Hendy also proposes the same idea for the Thessalonican issues of John III Vatatzes, where it is even less convincing.

Analysis of  Hendy's theories.

Hendy's theories have some superficial appeal, but, as we have already seen, on closer examination problems soon appear, so that we very possibly may need to modify some of his ideas.

At first sight the Series I types seem to constitute a reasonably homogeneous group, with average weights of around 2.3 gms, and with figures 18 to 20 mm high, while the Series II types, as originally defined by Hendy, are somewhat more compact, weighing maybe 1.8-1.9 gms on average, with figures 17-18 mm high. (With all the Latin period coins, the best indicator of the true module size is the height of the figures, or, as Jordanov prefers, with the John C-D types at least, the size of the dies^a. Both these factors vary much less than the weight and size of the flan, and in any case the larger module types are usually found clipped.  In this regard it is important to note that Bulgarian numismatists use "Series II" to mean something quite different to Hendy, namely larger examples of the Series III types; this is further discussed in the Appendix below).

The great bulk of the coins actually issued in John C-D's name are of course Series III types. The mean weight of these very small types, and also of their similar predecessors under Manuel C-D, is about 0.8 gm, barely half that of the contemporary Latin SM types, while the figures on the coins are mostly 10-13 mm high, and rarely more than 14 mm, as compared with the 13-15 mm typical of full height figures on the small Latin types.

This all sounds quite neat and tidy, but it should be remembered that examples of the Series I and II types are actually very rare, and in fact for most types only one or two unclipped examples are known (most of the published examples were originally collected in the Balkans by Bertele, and are now in the Dumbarton Oaks collection, and a few more are scattered through national and private collections). Therefore, in reality the mean weights of the various Series I and Series II types are not well established, and, although there does seem to be a significant variation in the weights of the various types, there is certainly no valid statistical basis for the splitting them into two separate but self contained series, as Hendy does. In fact, the unreality of the two series is evidenced by the confusion in the literature as to which of the Series some of the larger types should be consigned, and also by the fact that some types seem to straddle both series. All of this can be seen by plotting the known weights and figure sizes of the various types in decreasing order, which produces graphs which can easily be read as showing one group with a steady decrease in coin size, rather than as two groups of coins on essentially different weight scales.

Given this, and also the fact, as noted earlier, that the two (nominal) larger series don't seem to duplicate each other, i.e, the designs of the Series II types are different from those of Series I, it is reasonable to conclude that we are actually dealing with one combined series here, made up of coins of varying (mean) weights, rather than two separate series on distinct weight scales. As well, we note that there are also a number of types with sizes intermediate between those of Series II and III - such as the labarum head/canopy type S.2221 - which have a mean weight of 1.6-1.7gm and figures 15-16 mm high, a little larger, it should be noted, than the Latin SM types. For the sake of discussion, I will call these latter types "Series IIb" issues here. Note that, on the definitions used here, the "Series II" types actually pictured in Sear would seem to be closer to Series IIb than Series II.

(The Series IIb types evidently belong with Series II rather than Series III, since they seem to have smaller, if sometimes scarcer, Series III analogs. It may be suggested that these Series IIb's are just the larger end of the Series III weight range, but it does seem that we are dealing with two different weight scales, and we actually have somewhat better statistics to go on here than with the Series I and II types. Moreover, given that in some cases the known examples of Series IIb types outnumber their Series III analogs, it may be that in the end they largely superseded the very small types).

Overall, by my count there are now at least 18 different larger module types known, including two Type IIb's, and that's being very conservative, as I might well have added several more. (The Appendix below gives a complete listing of these larger types, with references to the actual evidence in the case of the newer, i.e, post Hendy, types).

To sum up, as far as the larger types are concerned, on the basis of the actual data  it seems reasonable to assume that we are dealing with a single series of types, although one with a fairly wide variation in weight. If so, this would mean of course that early in his reign John must have suddenly doubled or trebled the rate of issue of the larger types, to around one every four months or so. Why he did this is not immediately clear, but at least this idea makes as much sense as the parallel issues theory, which in any case says nothing about the Series IIb types.

But if we are dealing with only one series, how should we now arrange the various types, i.e, can we say anything about their order of issue?

The order of the types.

The most obvious way to order the larger types is by size, since we might reasonably assume that the range in size of the types was due to a decrease in their mean size over time. Hence, adopting for convenience the usual formal classification into Series I and II types, this would mean that the earliest issues were the Series I types, followed by the Series II and finally, the Series IIb issues. These types could then all be assigned to the period 1237-42, averaging about three issues a year, along with the corresponding "imitative" Series III types. Possibly the "novel" Series III types could then be assigned to 1242 and later, more or less as Hendy proposed, although once again there is very little hard evidence to support this last idea (more about the Series III types later).

This all seems fairly satisfactory, but one obvious problem with the arrangement is that we can't be sure that current "novel" types are really novel, since they can be instantly converted into an imitative type by the discovery of a single new larger module prototype. This is certainly something that we might well expect, particularly in the case of the commoner types of novel issues (such as the Cherub type S.2210, for example, which is in fact the second commonest Series III issue).

This arrangement may in fact finally give us a clue to the reason for the increase of the rate of issue of the LM types - the rapid turnover of types, and the decrease in module size to something close to the standard Latin SM size, may possibly represent an attempt by Thessalonica (separately from the Series III types) to accommodate itself to the clipping of the LM types which had become endemic in Bulgaria by the 1230's. In this regard it is also noticable that even John's Series I types are clearly lighter as a whole than Manuel's large module types.

The Series III types.

As noted earlier, Hendy wants to assign the Series III types to the Venetians, and to extend them into the 1250's, mainly because of firstly, the supposedly cruder and/or more imaginative style of some of the types, and secondly, the large number of types overall.

There is some force in this argument, but there are problems with this part of Hendy's theory as well. Firstly, most of the imitative Series III types are stylistically accurate copies of the LM originals, and hence are unlikely to have been produced by the Venetians, or anybody else, outside Thessalonica. Secondly, several of the more "imaginative" types appear in larger module form, (including some not noted in DOC IV, such as the B. Virg./Winged emp. type, and the Large Star/Virgin orans type - see Appendix for references to these last types), and hence, on Hendy's theory, these types would have to belong to the period of John's reign as emperor, i.e, before 1242.

Furthermore, why would the Venetians suddenly produce an ever changing galaxy of Series III types for use in trade anyway, when they were quite possibly still producing  small module later Latin Types T-W for precisely the same purpose? (see Article "The Bulgarian and Latin Imitative Types"). It  might possibly be argued that the new small types were meant to replace the small Latin types, but even if so why would they (the Venetians) suddenly halve the weight of their very successful small module Latin types?

In any case, whoever struck the Series III types, it seems clear from the hoard evidence that the series did not extend much, if at all, into the 1250's, as the Petrich hoard from southern Bulgaria, which was very likely laid down during the campaigns of John III in 1246-9 at the latest, includes almost all the known Series III issues, including the anonymous types. (And note also that the clearly later Dolna Kabda hoard contains no extra John C-D types - even the Lis/Pat. Cr. type had appeared in the earlier Serres (1960?) hoard, although this is not noted in the summary of the latter hoard in Jordanov).

Overall, in the end it seems to me that there is no hard evidence to support  Hendy's ideas on the Series III types, and hence that there are no compelling reasons for not assuming that these types were all issued, like the Series I and II types, in Thessalonica during the reign of John C-D as emperor. However, having said that, it's not impossible that some of the "novel" types belong to the  periods of John and perhaps Demetrius Comnenus-Ducas as despots, i.e, 1242-46.

Note that quite a number of Series III types are now known which are not included in DOC IV. The most notable of these are, firstly, a version of the Series I(?) Type D (S.2190), the only larger module type for which a small version was not previously known; secondly, a smaller version of the Latin SM Type A with the inscription "IwanicD" - a relatively common type at Petrich which has no doubt often been misidentified in finds and elsewhere as the small Latin A; and finally, the anonymous Lis/Pat. Cr. type mentioned earlier - this last is a simple, "makeshift" type, and very possibly a relatively late (post 1242?) issue, as the only hoards in which it is specifically reported are Serres and Petrich. (Note that there is an example of this last type in Jordanov (incorrectly called type T - cf. p.87 and Pl. 37-15) - unfortunately the origin of this particular coin is unspecified, but it does not seem to have come from a reported Bulgarian hoard).

Why the Series III, and why so many types?

In respect of the Series III types several basic questions still remain: - why were the small module Thessalonican types introduced in the first place, why were they struck on such a light weight scale, and why were so many different types issued under John C-D?

The answer to the first question is most likely that the small types were produced in Thessalonica for use in trade in Bulgaria, where the small module Latin types were the dominant coinage. The Thessalonicans were obviously receiving clipped examples of their own coinage back in trade from Bulgaria, so they revalidated them as small module types for use in Bulgaria, particularly under Manuel C-D (it's noticable that the small type C's of Manuel are often overstruck on clipped examples of the LM version of the same type).

The weight scale of the new types was then most likely determined by the weights of the cut down LM coins used for this purpose. Some of these coins were clipped down to the Latin SM standard of c.1.5 gm, but others were rather lighter, possibly because by the time of Manuel C-D they were often made by dividing the large module coins into three or four pieces, which after trimming average less than 1.0 gm in weight. These last coins seem to have become the model for the later small coins of Manuel C-D, and by the time of John C-D the very light weight scale was established as the norm for the SM issues in Thessalonica.

Anyway, whatever the original reasons for the new coins, they were evidently a success, at least for a while, as in the 1240's they became an effective alternative to the Venetians' small module Latin type A, and its successors the small types T-W, which for decades had dominated the Bulgarian trade. Coinage returns a profit to the mint as seignorage, so that under John C-D at least the Series III types may well have been issued, not just to facilitate trade, but also in order to capture some of the value of that trade (they seem to be found primarily in Bulgaria). The reason for the multiple small module types under John C-D may then have been, as some writers have suggested, to deliberately distinguish the new series from the old, unchanging Latin type. Or, as I personally like to think, it may simply have been part of an explosion of artistic creativity in the Thessalonican period, as witnessed in the fine style coins of the time - or perhaps it was both.

Note that we now have some interesting mules of Series III types. Since one particular coupling (of the large star obverse of S.2218 with the canopy reverse of S.2221) is known from three examples from different dies, it seems possible that different Series III types (at least) were struck simultaneously in the mint, and the coiners sometimes got the dies mixed up (although I have to admit that we don't see the reverse combination, so this may be deliberate combination). However, cf. also S. Bendall in N. Circ. '05, pp. 312-4 for an interesting brockage mule with the normal reverse of S.2205 and a brockage reverse of S.2222 on the obverse, and Goldberg Sale 55, Lot 596 for a mule of S.2209 and S.2210.*

One final point is that John III Vatatzes did not issue small module types in his own name after he returned to Thessalonica in 1246 in any quantity, except perhaps for Hendy's "Thessalonican" type K of John III (S.2134), which may have been issued in both Magnesia and Thessalonica (cf. "Sequencing the Latin Types"). As to the reason for this we can conjecture that trade with Bulgaria was no longer important by the late 1240's, since by then the Bulgarian economy had no doubt suffered greatly under the impact of the Mongol invasions. The Series III size types were therefore discontinued, as the Latin types (large and small) had already been.

*  Other possible mules and/or overstrikes are Bendall's Type Y (S.2211, DOC IV 25b), known only from the messy coin 499 of Bendall's reconstructed Latin period hoard in N. Chron. 1978, and Hendy's doubtful Type L (DOC IV 31b), known from - what? The "b's" seem to be a giveaway with these types.

Conclusion.

In summary we can say that:

   (1)  There is no real evidence to support Hendy's theory that the larger types of John Comnenus-Ducas were issued in two parallel series. (but see note on S.2337 in Appendix).
   (2)  There is no real evidence that the Series III types were struck anywhere except in Thessalonica.  
   (3)  There is no real evidence that any Series III types were struck after 1246, and in fact there is good (hoard) evidence that they did not postdate 1246 by much, if at all. 

Overall, if we adopt a consistent set of definitions for the various Series, we find that the coins of John Comnenus-Ducas can be divided into just two groups - the larger types, including Series I, II and IIb (which generally don't duplicate each other), and the small Series III types, some of which copy (all) the known larger types, and some of which seem to be original issues.
 

Appendix:    The larger John Comnenus-Ducas types.

The currently known* larger types are listed below by size, which may or may not correspond to the order in time. For compatibility with common usage I will use the standard classification of types into Series I, Series II, and my new Series IIb, although as explained earlier I think there is probably only one continuous series here, and it should also be kept in mind that types may be reclassified in future as further and better data on sizes and weights is obtained. Where the Hendy references have changed, the old (1969) one is given first, with the new (DOC IV) tag in brackets, i.e, I (Dii) means Hendy old Type I, new Type D (of, in this case, Series II). For unlisted large types the Sear No's of the corresponding Series III issues are also given, with the old and new Hendy/Bendall labels in lower case (to distinguish them from the larger module types), and with the DOC IV number in the footnotes. (On my analysis of course the relabelling of the Series II types by Grierson and Hendy is unnecessary and unjustified, and simply adds to the confusion. As well, the relabelling of the Series III types - apparently to match Grierson's ordering, which is followed by Bendall in Sear - is not only unjustified but also most undesirable, given the large number of hoard reports and journal articles which have used the old references. Unfortunately, this very confusing practice also occurs elsewhere in DOC IV).

In his article on the coins of John Comnenus-Ducas in Num. Chron. 2002, Bendall included a large number of new Series I and II types, based mainly on examples included in Dochev's book on the coins found at Turnovo^b. However, it should be noted that many (but not all) of the coins in Dochev's book are evidently shown up to 20% oversize, so that we have to assess his new Series I and II candidates with care, taking into account weight and reported actual size. Doing this we find that Bendall  followed (most likely inadvertantly) Dochev's light weight standard for Series II, which is similar to that of Jordanov (see below), but does not conform with Hendy's definitions. I have therefore made my own fairly conservative classification of the types here, and have omitted most of Dochev's Series II candidates (hence the differences between the table below and the similar one of Bendall). Nonetheless, the new table does include some additional Series I and II entries based solely on Dochev's examples; these seem to be valid, but at this stage they should still be treated as provisional.

I have also omitted for the present S.2217 and S.2222 - midweight examples of these are known (Bendall Num. Chron. 2002 Pl. 48,8 and Goldberg Sale 55-597 & 599), but the figure sizes are uncertain - and also a couple of "Series III" types in DOC IV which might well have been classsified as Series IIb. (These and the occasional smaller IIb types in the table below may be just overweight Series III issues, but they also are possibly issues on the Latin SM weight scale of c.1.5gm).

As well, all of Jordanov's own Series II candidates have been omitted, as the illustrated examples (in his book and in the Dolna Kabda hoard report) are all clearly Series III types. Jordanov takes Series II to mean coins struck from dies of 14-15 mm, with weights of 1.1-1.2 gm (Jordanov, p. 88), which are really part of Series III as originally defined and commonly understood. He also limits Series III types to dies of only 10-11 mm, with weights of typically 0.4-0.6 gms, which is far too restrictive. These definitions are based largely, one suspects, on his experience with the very late Dolna Kabda hoard, where almost all the coins are either heavily clipped or very small to begin with (also, Jordanov takes a readable obverse to be a marker of Series II as opposed to Series III - and note that the smaller coins in his book are mostly shown greater than real size, by varying factors).

For some more possible Series IIb and III types, see Coins 7-10  in the Article "The Coins of Michael II of Epirus". This rather odd type seems to be the only one which really does appear in in significant quantities in all three modules sizes, or least in a wide range of sizes.

Notice also the heavy weights (2.09 & 2.48 gm) of Lots 193.3-4 in the Despot Sale (LHS Numismatics Sale 97), which are listed there (but not illustrated) as DOC IV 17b, i.e, the Series III (new) Type Db - these may be the full length figure equivalents of the Series I Type E. (We should also perhaps note here S.2224, DOC IV-13, which is more a trachy of Series IIb size than a tetarteron).

Finally, the rare wing holding sword type, with a bearded ruler, (S.2226), seems to be of largeish Series III (i.e, Latin SM) size, and hence is presumably a late issue of John C-D, or just possibly, an issue of Demetrius C-D. Note that this type is apparently not known in a very small version.

*  Note that a number of new "Series I" versions of known Series III types have reportedly been unearthed during recent excavations in Ochrid, but I have not as yet seen a report detailing these new types.

References:

a:   I. Jordanov, "Coins and Coin Usage in Medieval Bulgaria, 1081-1261", Sofia, 1984. 
b:   K. Dochev, "Coins and Coin Usage in Turnovo, XII - XIV C.", 1990.  
c:   Odd example. Num. Chron.'02, Pl. 48,5 (Private Collection).
d:   Odd example. Petrich Fig. 193. 
e:   N. Chr.'02, Pl. 48,4, from Dochev Pl. 7,7. The Series III version is DOC IV-34.
f:    N. Chr.'02, Pl. 48,2 (Private Collection).  The Series III version is DOC IV-27b. 
g:   The Dochev/Jordanov coin is also shown in N. Chr.'02, Pl. 48,3. The Series III version is DOC IV-28. 
h:   Dolna Kabda 1624; CNG53-1959 (this last example weighed only 1.44gm, but the reverse figure as shown was 20 mm high). The Series III version is DOC IV-33. The larger size versions of this type are surprising, since otherwise we might have assumed that this anonymous type was a relatively late issue.
i:    For a recently found "Series I" (or Series II) example see Bendall, N. Circ. 2005, p.313. 
j:    For Series IIb size versions cf. Despot Sale LHS97-182.4 and Lanz Sale 141-903. 
k:   For Series IIb versions cf. Petrich Figs. 190,1.
l:    Series I in Hendy and DOC IV, but clearly smaller (a recent new example, Gorny & Mosch 181-2760, weighed 1.32gm with figures 16-17 mm high). For the unlisted Series III version (w. figs as busts), see Petrich 968,9; Despot Sale LHS97-183.2.  
m:   Listed as Series I (Type F) in DOC IV, but in fact it is Series II at most and.more likely IIb. (The St Achillee coin pictured in DOC IV is not shown to scale - the picture is adapted from the original site report where the coins are generally shown oversize). For other Series II/IIb versions cf. Petrich Figs. 190-2.
n:   For Series IIb versions cf. CNG 53-1913, Petrich 764, and Despot Sale LHS97-188,189. The (scarcer?) Series III version is DOC IV-36 - for other Series III examples see also Petrich 976,7. 
o:   For other Series IIb examples see Despot Sale, LHS97-192 & 193.13. For the (scarcer) Series III version, see Petrich 986.
p:   (Reserved).
 

Ross Glanfield

September 2007.

Latest revisions: 

  9 Sep. '07:   S.2223 added as Series IIb type.
  9 Sep. '07:   Discussion of two series theory revised.
18 Nov. '07:   Alternate theory on the ordering of the types dropped. 
28 Nov. '07:   Note on mules added.  
  1 Apr. '08:    Some of Dochev's new types in the table reclassified (crescent type dropped from Series IIb).  
25 Oct. '08:    S.2192 reassessed. 
25 Aug. '09:    New example of S.2190 from Gorny & Mosch noted.