06 January, 2012

Type held in the hand

Type, as we know, is something that you can pick up and hold in your hand. But there are fewer opportunities for doing this now than ever before, and our fading familiarity with the way type was made in the past can make for puzzles that are not easy to resolve.
Harry Carter’s phrase with which he began the lectures that became A view of early typography (1969), ‘Type is something that you can pick up and hold in your hand’, was a reproach to bibliographers whose attention was so occupied by the marks they saw on paper that they were not always much concerned about the means by which the solid types that made them were created and how this had affected the way they looked. The reproach was not always deserved, even when it was addressed to bibliographers isolated in their libraries. Making type was traditionally a secretive trade and difficult to penetrate.
In recently using a post to praise Talbot Baines Reed’s work as a historian I observed that his notes on the evidence for the shape of early types are still useful, and that they were retained with little alteration by his editor in the edition of 1952. But there is scarcely any reference in his original edition of 1887 to the way in which type was being made then – in what can now be seen to have been fateful years, when mechanization began to change the normal stance of the setter of texts from standing at a pair of cases to sitting at a keyboard.
It is often in my mind, when I use Reed’s book, that it would have been helpful if he had recorded more of the everyday operation of his typefoundry and added a note on the way it was changing. Notwithstanding some of the new techniques that had crept insidiously into regular use, like the use of the machines that had taken over the work of most of the casters with the hand mould and the electrotyping that had become a process for making many of the matrices, he must have been well aware that most of the operations of his own foundry had not changed in centuries. It still held one of the major surviving collections of early materials, the larger part of which were still usable, and the work of the casters and dressers had hardly changed. In the collective memory of the workmen there was a store of unwritten knowledge of how things had once been done.
Now we are well aware that not only has our own access to this tradition almost slipped away from us, but nobody today is quite sure, since metal types are not going to return, how we can preserve the embarrassingly rich stores of historical punches and matrices that have survived so far. The technical skills of those who understood the trades of the caster and the punchcutter, which were passed to the next generation by example and word of mouth and which would help us to understand exactly what we have left to us and how it was used, are now almost beyond our reach, even if we can find the individual sets of punches and matrices that we may be seeking among those that are stored.
I write this bit of sermonizing, not for the first time, to try to give some dignity to a post that might otherwise seem to deal with trivia. Its subject – some details of the way type looks, and why – is one that has been in my mind since I noticed that the typefounder’s mould in the image published in the Diderot Encyclopédie would actually make type with the nick ‘below’ the letter, just like modern type in English-speaking places, and not in the French manner.
The nick (see the image at the head of this post) tells the compositor whether all the type in the line is the right way up or not, and it distinguishes p from d in the case. The nick is visible to the compositor and it can be felt with the thumb, reassuringly, as the type is set. But French type usually has the nick ‘on top’, or at the back when it is placed in the composing stick. Fournier le jeune labours the point in a long section on the cran, or nick, in his Manuel typographique. Moreover, there is this image in the Encyclopédie, in the section on printing, that shows some large types which have their nick ‘on top’, in the French manner, as we would expect.

But what exactly does the Encyclopédie itself tell us? There is a whole article, a brief one, headed cran, which is included in Giles Barber’s admirable collection of selected materials from it on the making of books which was published in 1973. The piece is carefully worded, giving the function of the nick, but managing not to offer an opinion regarding which side of the types the nick should be. This (in my words) is what it says:
“Nick, a term of the typefounder, is a small indentation in the body of type which tells which way up it is. The compositor who takes care to place the nick of each piece of type on the same side is sure that they are all in the right orientation. The nick is on top or below, according to the country, and the choice of the printer.”
Fournier himself acknowledged that although the nick was ‘on top’ in France it was on the other side in Holland, Flanders, Germany and the ‘Lyonnais’ – the district round Lyon, the city that had been one of the major centres of printing in France. He says nothing about England, but then he had no knowledge of founders’ practice there. As Bodoni’s surviving moulds show, types in Italy had the nick ‘below’, like those in England.
Hence the puzzle of this diagram in the Encyclopédie, below. It is fig. 2 of plate II of the section on ‘typefounding’ (Fonderie des caracteres).

It shows the ‘bottom half’ of the mould, with the matrix pushed up to the ‘stool’ and its lower part projecting. The impression of the punch is not shown on the matrix as it should be, but there is no doubt that type it cast would have the nick ‘below’, in the English manner, where the compositor could see and feel it.
Even if the diagram is intended to show a kind of universal mould, with an orientation that would be familiar in many countries and not just one, it contradicts what one has learned of the French custom. Since the image among the engravings of the Encyclopédie that illustrate the article on printing (above) show big types with the nick on top, or at the back, is there a contradiction to resolve? What other evidence have we of the general practice in France?
We have in fact the advantage of an earlier treatise, the account of typefounding that was prepared by Jacques Jaugeon for the so-called ‘Description des Arts et Métiers’ with the authority of the Academy of Sciences. Its image of the mould, engraved by Louis Simonneau in 1694 (below), shows the nick on top. The closed mould is on the left, with the bottom end of the matrix, to which a piece of leather has been tied, projecting from it. The image in the centre shows the other end of the matrix, with a piece of cast type, with its nick ‘on top’ and the projecting ‘jet’ of surplus metal, still in place.

In fact the surviving iconography goes further. In the Bibliothèque de l’Arsenal there is a series of pen and wash sketches relating to typefounding that were probably drawn in the 1690s and intended for the engraver. They tell us more about the look of type than any other images of that date. Here is a large letter A.

Sure enough, the nick is ‘at the back’ in English terms. Moreover the projecting lower part of the type body has been shaved away at an angle with a plane, (‘bearded’ to use the English word), so that it would not pick up ink and print by accident.

Here is a view of the same letter seen from the side, showing the bearding clearly, and also something that the first sketch left out. The angles on either side of the A have been shaved, to eliminate the risk that these sharp projecting corners would pick up ink and print by accident. The feature that was left out has been added to both images. It is labelled in this one as the goutiere or ‘gutter’ (modern French gouttière), that is the ‘groove’ made with a plane in the foot of the type in order to remove the projection that may be left by the surplus metal that is broken off after it is cast, so that just two ‘feet’ are left, the feature (a brilliantly simple one) that gave each piece of cast type a uniform height. (The image above from the Encyclopédie shows these features too, though the bearding is not so well drawn.)
Details like bearding apart, we can be fairly sure that type of 1700 was not unlike the product of more recent techniques of casting and dressing, and that some of its details were innovations that were not features of very early types. It was nearly an inch high, as type seems always to have been – a nicely calculated size for something big enough to be picked up and assembled by hand and to lock itself into a solid forme. Here are types for text setting that were part of a purchase made in the 1670s by the University Press at Oxford with the help of the scholar Francis Junius, showing their nick and groove.

They are for a German Schwabacher for an ‘English’ body (about 14-point) equipped with special characters for printing northern languages like Swedish, which was the name that was given to the type at Oxford. Like all hand-cast types, these lack the precise finish of the machine-cast product, and their sides show scoring from being rubbed on a stone to remove the fins of metal that had crept round the edge of the hand-held mould. There is a conspicuous casting flaw in one sort, but if the face was good enough to print such a flaw in the body was of no consequence to the printer. It was said that the metal of hand-cast type was more solid than the machine-made product.
Here, for comparison, is some type that is only slightly later: an Etruscan type for the same body as the Swedish. It was made in London by William Caslon in about 1748 for sale to the Oxford University Press. The image below shows how it looks in the Caslon specimen of 1766. The type – almost the only type in existence from the original Caslon foundry – survived almost unused at Oxford because the opportunities for printing Etruscan were naturally so limited as to be virtually non-existent.

There is one other feature of cast type that the bibliographer sees and accepts without question, but which has caused more of a technical challenge to the founder (and irritation to the printer) than many readers can ever have suspected. Lower-case f has generally overhung the letter that follows: it is ‘kerned’ (to use the term in its original sense, one that has become modified in digital type-making). This was because f overhung the letter that followed in many early scripts, and typefounders, from Gutenberg onwards, have generally accepted the challenge.
In the 42-line Bible, a detail from which is shown below, as in most later types, f is kerned and hangs over the letter that follows. (In this example, it is also worth noting that there is a c that is cast to project beyond its body but in which the projection has been rubbed down flush to it, so that it abuts firmly on the specially-adapted form of i that follows, making a kind of ligature and keeping the rhythm of vertical strokes that contributes to the even visual texture of this type on the page.)

If the character following f was an i or an l, a special fi or fl sort needed to be made in order to avoid a collision; in several later types such collisions can be seen.
Here is one: the book, set in a nice bâtarde gothic, is Robert Gaguin, La mer de croniques et miroir historial de France, Paris, Nicole de la Barre c. 1520 (St Bride 5750). An f in the first line, in filz, has ridden up over the i next to it. Some five lines lower, a double-f ligature in suffisant has collided with the following i (or some other ascender) and sustained visible damage.

The kerned f became the common and inescapable feature of nearly all types, roman or gothic. (Perhaps to avoid the need for an fi character, Jenson put the dot of his i as far to the right of its body as it would go.) To give some idea of the engineering skills that were required to make the cantilevered projection of the f, here is the Canon or 36-point size of the so-called Fell type from Oxford:

Although precisely-engineered machines like the Monotype cast a kern that took some support from the shoulder of the following letter, founders generally shaved the underside of the kern in case variations in the height of the next letter induced stresses that could break the metal. In his Practical typecasting (1993), Theo Rehak, drawing on experience at ATF, specifies the use of the ‘kerning plow’, and also the use of a worn file for a substantial kern. Fig. 23 of Legros & Grant, Typographical printing-surfaces (1916) shows a ‘kerning file’, a device that held the individual type for kerning at the correct angle. One of these, from the foundry of Stevens, Shanks, was transferred to the St Bride Library.
In casting italics, the strokes of most of the descending letters like p needed to be kerned on the left and ascenders like l on the right, in order to get a good even rhythm. Italic f (together with long s) was a particular nuisance since it needed to be kerned on both sides: both kerned parts often tended to break or bend in use, however competent the caster had been. Eventually – but it took a long time – the habit of kerning f began to be lost. In the first decade of the 19th century Lord Stanhope, with his rational approach to everything and his enthusiasm for stereotyping saw no reason for keeping it (the plaster mould tended to catch in the kerned part of the letter). By the middle of the 19th century, when the long s had long been dropped from use, special ‘news founts’ with unkerned f were made for the printers of newspapers, who printed from stereotype plates.
Graham Pollard, a bibliographer who thought he had made a coup and identified a forgery by making much of its use of a kernless f apparently before its time, might have got himself into trouble if his antagonist, Thomas J. Wise (who was certainly guilty), had not been too old and tired to defend himself. Pollard had failed to look in specimens of the 1850s for kernless ‘news founts’.

As this image shows, the elimination of the kern from f did nothing to discourage the use of the fi and fl ligatured forms.
By the end of the 19th century, the practice of using non-kerning f for type for text setting was so widely accepted that for the makers of the Linotype machine, in which kerning was technically impossible, the limitation did not appear to be a significant handicap. Early designs made for the Monotype machine, which cast separate types, did not kern the f, perhaps principally to save trouble, since it was in fact possible. Eventually it was found by Monotype that kerned characters made italics that were better-spaced. Moreover this feature gave a conspicuous advantage over Linotype when traditional designs were made. And so the practice of kerning was reprieved, although in the last days of independent typefoundries the process was often handled with ever-decreasing understanding and skill. The italic of Monotype’s so-called Garamond type (series 156) handled the kerned characters very well: the long s is kerned even more generously than the f, demonstrating engineering skills that were no doubt those of the works manager F. H. Pierpont. The swash capitals are based on some made by Granjon for his Parangon italic, a type found in much printing in Paris and shown on the Berner specimen sheet of 1592. This is a specimen of Monotype’s 14-point Garamond italic. In this special version of the type, with its many archaic swash capitals and ligatured sorts, it seems likely that the long s was combined with some of the vowels that followed to make a single piece of type. Morison showed a synopsis of these characters on page 30 of his monograph On type faces, London, 1923.

But the machine could produce some impressive examples of kerning. The tail of the capital Q of Monotype Van Dijck (example below) runs right over the following letter.

For Linotype, for a long time the dominant system for typesetting in the United States, only a technical change would eventually overcome the limitations. In due course the liberating but flawed technology of filmsetting, and then its digital successors, notably OpenType, would do something to save the kerned f and to preserve the mandatory accompanying set of the five ff, fi, fl, ffi and ffl ligatures for painstaking and pedantic typographers.

French type and its nicks: a footnote
I wrote above that French type ‘usually’ has the nick on top. In fact I know of one exception to that rule, for a special reason.
The so-called Garamond type of the Imprimerie nationale has had much documenting in this blog (and of course in the magisterial Garamont website of the Ministry of Culture in France, introduced in October 2011). One odd result of the preserving of the matrices of Jean Jannon, acquired in 1641, is that as one can see that they have an unusual appearance : they are struck ‘upside down’. That is, the top of the letter is at the lower end of the matrix.
Traditionally the top of the letter in the matrix is at the end that is placed first in the mould. Making contact with the heurtoir, the ‘stool’ inside the mould, it sets the alignment of the type. The usual orientation of the letter on the matrix can be seen seen in this Figgins type of about 1810. The notch at the lower end is there to make it easier to tie a piece of leather to it so that it can be held in the mould, something that can be seen in the image of the mould by Simonneau:

The Jannon matrices were struck the other way round, so that the letter looks ‘upside down’. Perhaps it shows how new he was to typefounding. (As John Lane has noted, matrices for a very few English types of the 17th century were also struck in this manner, which perhaps tells us how isolated some English founders were from conventional practice.) Here is a matrix for Jannon’s Gros Canon (36-point) roman:

There is one result from Jannon’s habit that is perfectly logical but which surprised me at first. If type is cast from this matrix in a normal French mould the type itself will be ‘upside down’ on its body, so its nick will be ‘below’ the letter, as any English compositor would expect. This can be seen in examples of the 36-point Jannon type that have been cast at Ivry. I’m not sure that there is any moral to this anecdote, except that I find it reassuring that a kind of logical explanation can be found for such a minute variation from what seems like a rigidly undeviating and disciplined practice.

Note on nicks, etc.
Anyone at all interested in the discussion of this detail of the mould should look at Carter’s notes on its construction in ‘The history of the typefounder’s hand-mould’ in the Davis and Carter edition of Moxon, Mechanick exercises, revised edition, 1962, pp. 377–9. Kerning is dealt with pretty thoroughly by Fournier (Manuel typographique pp. 98 ff.; Carter, Fournier on typefounding, pp. 109 ff.).
An oversight. I am grateful to Stan Nelson for drawing my attention to the following passage, which I should have quoted in this post:
‘T. B. Reed noted (Old English Letter Foundries, 1887, p. 204) that the matrices for the Union Pearl, the Alexandrian Greek, the Court Hand, and the Scriptorials were struck with the letters upside-down. The same is true of Nicholls’s Great Primer Roman and Italic and the Great Primer Ethiopic at Oxford. If fitted in English moulds, therefore, they produce type with the nick at the back, which is normal in France. A possible explanation is that they were made for use with French moulds so as to cast type with the nick in front. Moulds of the kind appear, therefore, to have descended from Nicholls to Grover.’
Edward Rowe Mores, A dissertation upon English typographical founders edited by Harry Carter and Christopher Ricks,(London, 1961), p. 112. Horace Hart had already made some remarks on the orientation of the Nicholls matrices in his Notes on a century of typography at the University Press, Oxford (1900), at page 139.

Notes on some sources of images
I am grateful to those who have granted access to the materials shown here.
The image of the mould engraved by Louis Simonneau in 1694 is one of a series of plates illustrating typefounding that were prepared but never published, and copies are almost unfindable outside Paris. (But it is known that some proofs escaped. Copies of these may possibly exist in libraries or print collections, perhaps, since they have no captions, mistakenly identified as plates made for the Encyclopédie.) I showed these images with a commentary for the first time in an article, ‘Illustrations of typefounding engraved for the Description des Arts et Métiers of the Académie Royale des Sciences, 1694 to c. 1700’, Matrix 11 (1991), pp. 60–80. They match the manuscript description of punchcutting and typefounding compiled by Jacques Jaugeon for the Description des Arts et Métiers in the Bibliothèque de l'Institut de France, Paris. Fred Smeijers later showed them in his book Counterpunch (1996). The originals were from an album at the St Bride Library, London. The printed image of the mould is missing from this set of plates but the original copper plate had been acquired by the Newberry Library, Chicago, which was kind enough to have a few new impressions made from it, from one of which this detail is taken.
The two details of sketches showing a printing type for capital A appear on leaf 80 of an album relating to the Description des Arts et Métiers (Gr. fol. 114) at the Bibliothèque de l’Arsenal, Paris.
The early types from the University Press, Oxford, are among materials that were transferred to the St Bride Library in 1989.
The matrix of the Gros Canon roman of Jean Jannon is in the Cabinet des Poinçons of the Atelier du livre d’art et de l’estampe of the Imprimerie nationale, Ivry-sur-Seine.
The detail of the word faciem from the 42-line Bible (Leviticus 20:3) is from a single leaf at the library of Trinity College, Cambridge, one of several from a broken copy that were sold in New York in 1921 with a common title-page reading A noble fragment, being a leaf from the Gutenberg Bible.
There are selected articles from the Encyclopédie in Giles Barber’s Bookmaking in Diderot’s Encyclopédie: a facsimile reproduction of articles and plates, with an introduction (Farnborough: Gregg, 1973)