19 March, 2007

Big brass matrices: a mystery resolved?

Above. A brass matrix for one of the sets of titling capitals of the romain du roi, Imprimerie nationale, Paris. Photograph made in 1993. Letter height 31 mm.

In a note to the reprint of Moxon’s Mechanick exercises (2nd ed. 1962, p. 154), Harry Carter wrote, ‘it is a mystery to find matrices of brass struck with big punches in the sixteenth century at the Plantin-Moretus Museum and in the seventeenth century at Oxford.’
The reason for Carter’s puzzlement is as follows. Punches were normally cut in steel and hardened so that they could be hammered into copper in order to make the matrices from which type was cast. Copper, which is ductile, is a good metal for the purpose. Brass, which is generally an alloy of copper and zinc, is far harder. (It was used in the 20th century for the matrices of Linotype and Monotype machines, but mostly in relatively small sizes, and the punches were struck using powerful modern machinery.) So how is it that several sets of surviving early matrices for quite big types, for which the punches were cut in steel, are all made of brass?
The two examples that Carter had in mind were probably these. At Antwerp there are the unfinished Grosses capitales extraordinaires of Claude Garamont, bought by Christophe Plantin after his death in 1561, for which there are steel punches, brass matrices, and also a set of strikes in lead. The height of the letters is 14 mm. And at the University Press, Oxford, there are 38 brass matrices for the 3-line Pica titling capitals of the ‘Fell’ types, 11 mm high. ‘They cannot have been made by striking—the metal is too hard,’ wrote Carter in his notes to Stanley Morison’s monograph John Fell, the University Press, and the ‘Fell’ types (1967), p. 150. ‘The letters must have been cut out of the brass with chisels and finished by striking with the punches whilst the brass was hot (the method recommended by Fournier le jeune in his Manuel typographique).’ The term used by Fournier, on pages 75 and 76 of his first volume, is cuivre or copper, and although as noted below cuivre can mean brass as well as copper at this date, the process he describes is so laborious that it is difficult to believe that a hard metal like brass would be used for it.
There are more surviving examples of early matrices for big types made in brass.
When he was in Paris as the Minister of the United States, Benjamin Franklin bought some matrices dating from about 1740 of the foundry of Claude Mozet. The surviving matrices, for a big titling type (measuring 16 mm), are of brass. They belong to the Massachusetts Historical Society in Boston, were recently shown in the Franklin tercentenary exhibition in Philadelphia, and are illustrated at page 244 in the article by Ellen R. Cohn, ‘The printer at Passy’, in Page Talbott (ed.), Benjamin Franklin in search of a better world (New Haven, London: Yale University Press, 2005), pp. 235–71.
At Haarlem, among the materials in the museum of Joh. Enschedé en Zonen, there is a set of brass matrices for a titling type of the mid-16th century, probably French, measuring 16 mm. There are no surviving punches of steel, but there is a set of strikes of the same letters in lead.
And then there is the series of ‘two-line’ or titling capitals for the romain du roi among the materials of the Imprimerie nationale, Paris, cut during the 18th century. There are steel punches for these, but all the matrices, for types of which the largest size has a face 41 mm high, are brass. They could not possibly have been struck with these punches.
In this last case we have an answer to the puzzle, and one that seems both simple and convincing. In his accounts for the year 1728, asking for payment for one of these big types, the punchcutter Jean Alexandre claimed 90 livres,
Pour avoir fait soixante grosses Matrices de plomb frapées avec les gros poinçons des grandes lettres de deux points du quatorzieme, pour en pouvoir tirer les creux, pour avoir lieu de les mouler avant que de justifier les Matrices en Cuivre: a raison de 1 [livre tournois]: 10 [sous] pour chacune matrice.
This is not easy to interpret in every detail: why, for example, were 60 matrices were needed for a set of capitals? But it seems to mean something like this:
“For having made sixty big matrices in lead, struck with the big punches for the large two-line letters of the Fourteenth [body], in order to make impressions with which to cast brass matrices...”
Cuivre is an ambiguous word in French at this period, since it can mean either ‘copper’ or ‘brass’ according to the context. But in this case it seems more likely that it means ‘brass’ because the surviving matrices for these big types are all of brass. ‘Fourteenth’ is a reference to the scale of interrelated type bodies invented for the romain du roi by Sébastien Truchet, an invention for which the credit was disingenuously claimed by Fournier le jeune. The capitals for the two-line letters for this body are about 25 mm high or about 72 points. Thus the lead strikes that are referred to in the accounts were probably patterns that were used for making replicas in brass by casting in sand. When these castings were made into matrices, the faces of the letters would be cleaned up by striking with with the steel punches that had been used to create the impressions in lead, a practice sometimes followed in making medals: in the article on ‘Coining’ in the 2nd edition of Chambers’s Cyclopaedia, 1738, it is noted that ‘medallions, and medals of high relievo, by reason of the difficulty of stamping them in the balancier, or press, are usually first cast or molded in sand, like other works of that kind, and are only put into the balancier to perfect them; by reason the sand does not leave them clean, smooth, and accurate enough.’ That these matrices were cast in sand and then cleaned up with the original steel punches is the conclusion of Stan Nelson, former typecaster at the Smithsonian Institution and now a maker of typefounders’ moulds and the writer of their history. He believes that the matrix for the ‘Fifteenth’ body (31 mm) of the titling capitals of the romain du roi, shown in the photograph at the head of this post, shows evidence of having been cast in sand, and that some traces of a casting process are also visible in the Mozet matrices bought by Franklin.
So it seems not only possible but even quite likely that the sets of brass matrices listed here were cast in sand, and that the lead ‘strikes’ that accompany some of them were patterns used in the casting process. One should add that, since he fails to mention it, Fournier had either never heard of this process or did not believe in it. It should also be borne in mind that punches for larger-bodied types were sometimes cut in brass and used to make lead matrices, from which – with great care – type could be cast. It is known that this was done in Germany, and there are surviving examples of such leaden matrices in the Norstedt Collection in Sweden. But it seems highly unlikely that a set of lead matrices that were primarily intended for casting type would be associated with steel punches and brass matrices for the same typeface.

With technical advice from Stan Nelson – to whom I am also grateful for help with the present text – I first put a brief note on the accounts from Alexandre and the brass matrices of the romain du roi at the Imprimerie nationale into Le romain du roi: la typographie au service de l’État, the catalogue of the exhibition held at the Musée de l’imprimerie, Lyon, in 2002. (It is note 73 on page 64, and there are more images of the matrices for the 15th body on page 51 and 67.) The image at the head of this post was made in 1993 on Kodachrome film. The digital image just above of the same matrix was made in 2008. The cancelled figure 15 is a reference to the 15th body in the scale for the bodies of the romain du roi devised by Sébastien Truchet. The figure 48 which replaces it, which may have been added in about 1810, gives the size of the type for which it is designed to serve as a ‘two-line’ initial in the point IN, the ‘point millimétrique’ of approximately 0.4 mm that is still in use at the Imprimerie nationale. The type body is therefore 96 points. The image below shows a punch for the body one size larger, 112 points.