This article is due to be published in the Catalogue of Glass in the Department of Greek and Roman Studies at the British Museum.
Since the publication of Apsley Pellatt's book1, several theories regarding the manufacture of Roman ribbed bowls have been advanced. Some have suggested that the ribs were formed by stamping a disc of molten glass using a flat, circular mould with radial grooves2, before slumping it over a form, whilst others have proposed that ribs may be raised by flattening an already slumped glass bowl using a metal bar3.
1Apsley Pellatt (1849) "Curiosities of Glassmaking" (London: David Bogue) pp.104 -106
2Cummings, K. (1980) "The Technique of Glass Forming" pp.26-29 (London: B. T. Batsford Ltd)
Tait, H. (ed.) (1991) "Five Thousand Years of Glass" (London: British Museum Press) p.222, figs. 60 -64
3Stern, E. M. & Schlick - Nolte, B. (1994) "Frühes Glas der alten Welt. Sammlung Ernesto Wolff (Stuttgart) pp. 76 - 78
Lierke, R. (1993) "Aliud Torno Teritur - Rippenschalen und die Spuren einer unbekannten Glastechnologie: Heißes Glas auf de Töpferscheibe", in Antike Welt 24,3, pp.218 - 234
Lierke, R. (1999) "Antike Glastöpferie" (Mainz: Philip von Zabern) figs.124 - 128
Pellatt's description of these bowls as 'pillar moulded' has, unfortunately, given the misleading impression that a mould or stamper was essential, whereas we have found the relatively simple process outlined below to be most successful. The work we do always requires a certain amount of experimentation, and by far the most important conclusion that we have been able to draw from our experiments is that the early glassmakers invariably made use of the quickest and simplest methods available. We believe that their ribbed bowl production techniques are no exception to this rule.
In brief, our technique is a simple, relatively fast and repeatable method: Using a glass virtually indistinguishable from typical glass compositions in use at that time, each radial rib is individually formed on a hot, flat glass disc, using a pair of specially adapted pincers4. This tool resembles a pair of simple U-shaped shears, each blade of which has been twisted through ninety degrees so that both are parallel and pointing downwards. The glassmaker deftly pinches seven or eight ribs in quick succession before re-heating is necessary, and after three or four such 'passes' the hand-ribbed disc is slumped or sagged over a solid form. (All of this is performed 'live' at the mouth of a small, dedicated furnace). After annealing, the bowl is ground to remove the debris and toolmarks from the manufacturing process and polished.
1The use of pincers for rib forming has been suggested by at least two previous writers:
W. B. Honey (1946), p.23, writes "The dishes and bowls…....were commonly decorated with raised ribs (the so-called 'pillar moulding'), produced apparently by working up the plastic glass with pincers and other tools…"
D. F. Grose (1974) "Roman Glass of the First Century A. D. A dated deposit of Glassware from Cosa, Italy" in Annales de l'AIHV 6, p.37, fragment CE 2155, writes "Slight unevenness on interior surface opposite ribs may indicate that the ……. ribs were pinched out from the sides."
We have investigated the use of moulds and stampers in various different materials available to the Romans in some detail, and we have concluded that they are unlikely have been used to form the ribs. Our reasons can be summarised as follows:
Had the Roman glassmakers made moulds or stamps (for the pressing of the flat glass disc before sagging over a form), it could be argued that the mould maker might have taken the trouble to create one that was neat, well-made and symmetrical, with evenly spaced ribs of equal length, height and shape. It is also likely that before carving such a mould he could have divided the surface into an equal and even number of divisions for the ribs, 12, 16, 24 etc., since this is a straightforward and easy way to divide a circle. No evidence of such repeated patterns of directly opposing ribs has been discovered on extant bowls or fragments; instead, many bowls display an odd number of ribs of different lengths, heights, and spacings.
To our knowledge, no two ribbed bowls have ever been shown to originate from the same mould. (By way of comparison, many examples of mould-blown circus beakers which have definitely been blown from the same moulds have survived, yet these vessels are much less common than ribbed bowls).
When experimenting with the creation of a stamp, we found that establishing a suitable material proved a problem. Wood burns and becomes unusable very quickly when brought into contact with molten glass. Stone (such as sandstone, known to have been used for Roman prismatic bottle moulds) can crack when subjected to the intense heat necessary to press a molten glass disc successfully. Experiments with fired clay and plaster stamps yielded similar problems to stone, and the plaster rapidly became friable. Using a metal stamp created rippled cooling marks which could only be removed by re-heating the impressed disc to such an extent that the ribs also melted back into the glass. (In fact, all mould materials tend to leave some impression on a glass disc, and none can be detected on the surfaces of original vessels).
Using separators (such as powdered clay) was also less than successful; the clay tended to stick (and fire) to the molten glass and impart patterns. When water is used as a separator (with wood, stone, ceramic or plaster moulds), excessive steam is created, which can prevent the glass from completely filling the depth of the mould ribs.
Finally, the use of a mould does not explain the presence of the various idiosyncrasies seen on these bowls, which include: toolmarks along the lengths of the ribs where they meet the body of the bowl; the pronounced mound sometimes seen between the ribs (which we call the 'pillow' effect); a slight serpentine twist to the rib pattern; the evidence of some form of pressing and flattening at the wider end of the ribs, and toolmarks on the area between the tops of the ribs and the rim.
The published results of experiments undertaken to illustrate the use of the bar-forming technique5 give the impression of somewhat thick-walled, heavy bowls with disproportionately large ribs. The spaces between the ribs appear to be flat and do not exhibit the convex mound typical of Roman bowls. For reasons of economy, one of the aims of the glassworker should be to produce a ribbed bowl which needs as little grinding as possible, and this method does not appear to achieve this. The use of a rib pinching technique provides straightforward answers to the many questions concerning ribbed bowl manufacture which mould-pressing and bar-forming do not.
5Lierke, R. (1993) "Aliud Torno Teritur - Rippenschalen und die Spuren einer unbekannten Glastechnologie: Hei*es Glas auf der Töpferscheibe", in Antike Welt 24,3, p.231, figs. 32f - 32h
In our method, which we outline in detail below, each rib is individually squeezed on a flat molten glass disc. This results in slightly unevenly spaced ribs of differing lengths, heights and shape. (These idiosyncrasies diminish as the glassmaker's skill and experience increases, but never entirely disappear). As the spaces between the ribs have no contact with a tool whatsoever, there are no visible cooling or tooling marks in these areas. The 'pillow' effect seen on some bowls is created as a direct consequence of the pincers pressing evenly into the glass before squeezing. Most toolmarks formed by the pincers are smoothed during re-heating, resulting in a disc that is fire-polished on the outside. If the pinched ribs are allowed to extend to the rim, an undulating 'batwing' effect is created, which, although it is lessened by the flattening technique, and can be entirely removed by grinding, is best avoided from the outset. We have found that it is preferable to begin to pinch ribs at about a centimetre from the rim of the molten disc, and to allow the edges of the bowl to sag all the way to the ceramic batt upon which the form stands, which helps to give a consistent, flat rim to the vessel.
To form and sag a bowl using this process takes us approximately fifteen to twenty minutes. Molten glass is poured from a gathering iron or crucible onto a marver which has been dusted with dry powdered clay. It is flattened with a large, damp block of wood to form a disc, and transferred onto a ceramic kiln batt for re-heating. Our marver is made from a kiln batt, which helps to retain the heat of the glass disc (unlike a modern metal marver, which will quickly conduct much of the heat from the underside of the glass), and a thinner disc results. The underside of the molten glass disc picks up some of the powdered clay, and acts as a separator during the next phases.
The flat disc is transferred onto another, preheated kiln batt which is securely seated on one end of a long metal pole. This makes it possible to re-heat the disc whenever necessary.
Working quickly with the pincers, about seven or eight ribs can be made before the glass becomes too stiff and requires re-heating. (There are examples of original bowls with over twenty, or even thirty ribs). As the number of ribs pinched grows, the length of time during which those already made are exposed to the heat will also increase. (This can result in a certain amount of re-melting of these earlier pinched ribs, an effect sometimes seen on original vessels). If necessary, once a complete series of ribs has been pinched, they can be reinforced by pinching for a second time. This second pass is relatively fast, as the raised ribs are easily gripped. As the ribs are now slightly cooler than the rest of the disc, they are less inclined to re-melt during the subsequent slumping phase, which may be performed at a lower temperature in order to preserve definition. This action can also reinforce the slight serpentine shape often seen on the ribs of original ribbed bowls (as seen on catalogue nos. 533 and 579) caused when the disc is caught and repeatedly moved by the pincers, forcing them to distort.
It is important to ensure that the ribs do not become too hot and remelt during the pinching process. We achieve this by re-heating only that part of the disc which is to be pinched in the next 'pass'. This is usually the front portion of the disc - the part closest to the glory hole. We use a similar technique when making matt-glossy, or 'cast' window panes.
The number of ribs, the 'pillow' effect and the thickness of the glass between the ribs are all directly related. To maximise the number of ribs whilst keeping the other two effects to a minimum, it is important to pinch each new rib as close to the previous rib as possible. This also has the effect of making the ribs slope in the same direction as they are being pinched, and causes the thickness of the glass in the spaces between the ribs (when seen in cross-section) to taper slightly - an effect seen on original vessels.
The narrow gaps between the ribs widen during slumping. The circularity of the disc and its diameter can be adjusted during and immediately after the pinching process using the rib pincers.
After the ribs have been pinched, the disc is placed on top of the pre-heated form, re-heated in the glory hole and slumped. This requires care and concentration to avoid folding of the disc, and overheating, which will distort and thin the glass. The softened disc is coaxed over the form using metal or wooden tools, and allowed to run down the side of the form to meet the kiln batt, creating an even rim. Any areas which do not meet the batt can be teased downwards with a metal blade. This results in horizontal or nearly horizontal linear indentations which mirror those on some original bowls.
The material from which the forms were made deserves mention here. We make our forms from highly refractory ceramic, which enables us to re-use them many times, and create bowls consistent in size and shape, and we remove the bowl from the form before annealing. Such consistency is not seen in original cast bowls, however, and one possible explanation may be that Roman forms were designed to be discarded after each use.
It is possible that some ancient forms were made from a similar friable material to that known to have been used for core-formed vessels. There are several ribbed bowls with constrictions below their rims which cannot have been formed by grinding after cooling. The bubble map of catalogue no. 539 lends support to this theory: the single chevron pattern formed by the two bubbles is likely have been caused by a tool being employed in a simultaneous pressing and dragging motion, pushing the glass into the constriction. The form would need to be broken and discarded after the bowl was made, since a vessel cannot be removed from a solid form with such a re-entrant profile. It would also have to stay in place through the annealing and cooling process, and modern experiments show that core-forming materials can be made which allow this by matching their rate of shrinkage on cooling to that of the glass6.
6Gudenrath, W. (1991) "Techniques of Glassmaking and Decoration": pp.214-215 - Vessel core-forming in Tait, H. (ed.) (1991) "Five Thousand Years of Glass" British Museum Press: London
Labino, D. (1966) 'The Egyptian Sand-Core Technique: A New interpretation' in Journal of Glass Studies, 1966, pp.124-127. Note 4, p.125: 'It (the core-forming material) must have a coefficient of expansion greater than that of the glass, to prevent the glass from cracking upon contracting'
Following slumping, the region immediately below the rim of a ribbed bowl needs to be flattened in order to form the flat extremities of the ribs, and to lessen any toolmarks made during the rib pinching. We do this by re-heating the bowl and pressing a tool such as a narrow metal bar (or the flat handle of the pincers) against it. Several re-heats are necessary in order to complete this operation. The resulting folds and the steep, sometimes flattened slopes at the tops of the ribs are the same as those seen on original vessels (catalogue nos. 590, 591 and 650).
After annealing and cooling, the bowl has to undergo cold glass working to remove all the partially fused deposits of clay separator from the inside of the vessel, and to remove any toolmarks on the outside of the vessel, the removal of which may account for the quite heavily ground rims found on original vessels (catalogue no. 592).
This is done by grinding and smoothing the top of the rim and the outer portion below the rim using vertically-mounted grinding and polishing wheels of fine-grained sandstone before working the inside. This interior grinding can be done by mounting the bowl in clay which is allowed to become leather hard before securing to a powered turntable. As the bowl is rotated, and using traditional grinding and polishing media such as corundum, sandstone and pumice, the inside of the bowl can be ground and polished. If desired, and using the same grinding media (but this time formed to a sharp point), concentric rings can be cut as decoration to the insides of the bowls.
Rib pinching as a technique was not confined to these vessels. Some of the earlier free-blown vessels, notably small trailed bowls (often known as 'Zarte Rippenschalen'), exhibit ribs which are clearly individually pinched in a similar manner to those of non-blown vessels.
Many large storage jars, tall conical jugs, spherical and biconical jugs and small flasks all have hand-pinched ribs, which are not mould-blown as sometimes stated.
In conclusion, we believe that rib-forming using pincers is part of the mainstream of Roman glassmaking, and connects ribbed bowl making directly with the traditions of mosaic glass and 'cast' window glass, both of which are made and tooled 'live' as a molten disc or rectangle. Ribbed bowls, many of which predate the invention or discovery of glassblowing thus form a strong connection with early free-blown vessels.
Mark Taylor and David Hill 2001 (revised 2003)