Results

The find context: a reminder

This chapter presents a reconstruction of the production process of two horse statues on the basis of the traces of manufacture. The remains of the two horses were found together with fragments of the riders. This cache of around 1460 fragments in all was discovered in 1961 in Insula 28, where it had been hidden in the area of a disused sewer by the outer wall of a residential building around the middle of the 3rd century AD. The careful layout of the pit, as well as the waste from casting and two fragments of large bronzes found in the adjacent house, suggests that this was no hastily hidden metal hoard, but rather a recycling store that had been deliberately laid out and subsequently used in conjunction with a neighbouring bronze foundry.

The differences in manufacturing technique between the two horse statues

Of the 368 fragments of the «scrap find» («Schrottfund») which definitely come from Horse I, 149 could be joined together into 17 segments (40.9%); of the 445 fragments of Horse II, 170 fitted together into 21 segments (38.2%). Thus, for the reconstruction of the two horse statues, a good third of the fragments were used which could be attributed either to Horse I or Horse II on the basis of their characteristics. Located on one or other horse's body by means of anatomical criteria, for the moment being, the segments are the point of departure for reconstructing as completely as possible the two statues and their individual manufacturing processes. The basis for this was the systematic recording of the traces of manufacture- to be found primarily on the inner surfaces - as well as their documentation in technical drawings and in the catalogue. With increasing knowledge, the traces of working have proved to be criteria to be taken seriously for or against the locating of individual segments. They made it possible to reconstruct the manufacturing process of each horse statue phase by phase, the specific differences showing up clearly by comparison.

Already the subdivision of the two statues into separately cast casting sections reflects two basically different approaches to the casting of statues. Ten casting sections can be presumed for Horse II , of which seven are evident in the surviving segments. Horse I, on the other hand, was constructed from at least 25 casting sections, of which 19 are evident in the located segments. Of all the casting sections of which the format can be reconstructed, those of the rump are especially informative in terms of the general characteristics of the manufacturing process. While the torso of Horse II is constucted from just three casting sections, that of Horse I is put together from ten such sections. The advantage of this subdivision into small elements is firstly, no doubt, that the sections could be cast more thinly, with a more economical use of materials. Secondly, small casts are less risky and can be repeated as necessary with greater ease. This can be proved to have happened in two cases.

The size of the casting sections also determines their form. The large casting sections of Horse II can be described as bell-shaped, since the opening of the hollow body is always at the widest point. The small casting sections of Horse I, in contrast, can be characterised figuratively as «relief panels». Only the lower halves of the legs are tubular casts, as those of Horse II.

The differing format of the casting sections of Horse II and Horse I necessarily presupposes a different use of the same moulding process. The traces of manufacture which have remained in the positive on the inner surface of the fragments indicate that in both cases the wax models for the casting sections - which, taken from the original model, are in the negative - were made by the technique of indirect moulding (by use of an «inter-model»). In this process, several negative mould-sections were used, which were each divided into at least two parts. With Horse II, the course of the wax seams between the wax slabs and the moulded section edges indicate that the format of the wax model is identical to the format of the later casting section. For Horse I, by contrast, the course of the surviving wax seams provides evidence for the existence of section edges which have been moulded on the one hand and cut on the other. The wax models which derived from negative mould-sections were thus divided, once out of the mould, into small models for casting the mainly slab-shaped casting sections. This must have been done on a scaffolding, the struts of which have left impressions on the inner surface and the rectangular openings in the walls.

Because of their format, all the casting sections of Horse II must have been cast with a core. This must have been introduced into the wax model when it was still embedded in the negative mould-section. In the case of the casting sections for the tail, the core was filled up at the time that the tail was being moulded. With the smaller casting sections (legs, tail), it can be assumed that the clay of the core was poured in while in a viscous state. With the large casting sections, however, impressions on the inner surface indicate that a previously made, reinforced core was pushed through the large opening in the neck or the back and that only the hollow space between the core and the wax wall was filled up by pouring in liquid clay. This is confirmed by the holes left in the cast walling made by the spacers , the raised edges of which are caused by the insertion of the iron rods pushing the soft wax wall inwards. With the slab-shaped casting sections of Horse I, on the other hand, there was no need for a core. The positions of the spacers and the run-in channels support this view. Here the spacers were placed, in contrast to Horse II, only in the area of the section edges, and the run-in channels are positioned both on the inner and the outer surfaces.

Technical difficulties in casting can arise from the format of the casting sections of Horse I and the thinness of their walls, on average 1-3 mm. In order to ensure an optimal distribution of the molten bronze flowing into the mould, casting aids were employed for several casting sections.These feeders present themselves as bulges on the inner surface. During casting these formed a system of channels to lead the inflowing bronze to the more inaccessible places within the mould. These have survived on three casting sections: a massive penis on the underside of the belly, on the narrow, raised chest, and on the croup, which was cast as part of the side of the belly.

Because the joints between the large casting sections of Horse II had to withstand enormous pressure, they were made very carefully either from single casts closely abutting or overlapping each other (joint type 1a), or from multiple casts (joint type 1b). Since the actual joints consist of the welding of the section edges with the added bronze, half circles were cut out of the edges at opposing points and the edges thus lengthened. In Horse I, single casts (joint type 2a) and multiple casts (joint type 2b) also join the individual casting sections by cast-welded. Here, however, the section edges form irregular rectangular openings to receive the added bronze, which moreover was only applied at intervals. Between these points, the join between the abutting section edges was covered with patches. This technique of joining sections with punctual cast-welding, used as a rule in Horse I, again takes into account the division of the statue into numerous small casting sections, because here, in contrast to Horse II, the static pressure within the statue is distributed over many more seams. Only in mounting the stepping back legs to the rump was the joint made, for static reasons, using a massive collar cast (joint type 3). Overcast joints were used in Horse II only under the shoulder, for the free hanging casting sections of the two raised front legs (joint type 1c), and in Horse I for the casting sections of the tail as well as of the neck at the lower edge (joint type 2c).

As already with the wax model, there is also evidence for the use of a scaffold when joining the casting sections of Horse I, the rods of which were poured into openings cut into the casting. Because the holes remaining after the removal of the rods were closed with patches, these could not have been used for any possible framework for the finished statue.

Basic differences in the manufacturing techniques of the two horse statues are also evident in the finishing process. While in Horse II casting faults which penetrated through the walling underwent the separate step of repair-casting, the numerous casting faults and surface imperfections in Horse I were repaired exclusively by hammering in patches during the process of engraving. Because these were also soldered with tin from the inside, large areas of the inner surface were smoothed with scrapers and files. In contrast, this sort of patching was used very seldomly in Horse II, at most in cases of large surface damage. Instead, when damage occurred as a result of engraving, as for example with burst blisters, this was mended by hammering in small pieces of bronze.

The anatomical details in Horse II were very carefully recut with a cross-cut chisel and engraved with a burin. This work was carried out in a summary fashion in Horse I, although the recutting of patches in the area of folds in the skin indicates a careful and practiced hand.

Perspectives

The comparative description of the manufacturing techniques of Horse II and Horse I from Insula 28 in Augst brings to light differences at all stages of their production, which are due to their manufacture in different workshops (table 7). Moreover, two basically different approaches to the casting of statues is evident in this: while in Horse II the division into casting sections indicates its conception as a plastic structure, Horse I - to exaggerate - is a piece which has been put together from «relief panels». Correspondingly, the traces of working which can be related to the individual phases of manufacture in Horse I indicate a careful but considerably more routine process than in Horse II.

This can be understood as evidence of the emergence of a mass-production of Roman bronze statues, increasingly reflecting economical considerations. The cause of this can be found in the contemporary custom of erecting likenesses not only of the emperor and numerous officials, but also of any other worthy citizens. This is best documented in the literary sources for the city of Rome itself. According to the sources, this was practiced already to such an extent in the early imperial period that, for example, Emperor Claudius had most of the statues «taken elsewhere»: and he forbade every private individual in future from «continuing this exercise, unless the Senate granted him special permission or unless the person concerned had had a public building erected or renovated». Before him, Augustus had already had the statues of famous men moved from the Capitol to the Field of Mars because of lack of space. And for the late Roman period a list enumerates for Rome alone 2 colossal statues, 22 colossal equestrian statues and 3785 bronze statues of emperors and army commanders. In the 6th century AD Cassiodorus remarked that a second population of statues seemed to live within the walls of Rome. There also appears to have been a similar situation in the provinces of the Roman Empire, for on the forum of Timgad there are still 58 bases for bronze statues in evidence, while on the forum of Djemila there are a total of 63.

If one puts the present study into relation with the written and material evidence of Roman mass production of large bronzes, the agreement established in several instances of the manufacturing techniques becomes more significant: Horse II, on the one hand, with the early imperial horses of San Marco, and Horse I, on the other hand, with the horse of Marcus Aurelius and the head of Constantine. It would follow that the routine production of Horse I, as compared with Horse II, may be characterised as a development of the production technique which arose because of the growing market demand. This thesis may serve as a spur to studying the manufacturing techniques of classical bronze statues further in terms of the Roman bronze industry.

(Translation: Catherine Aitken)