The Art of Founding Type

By John Downer

This text was first published in 1995 on the back of a poster introducing the Not Caslon typeface.


Caslon types have been in existence now for about half as long as the art of typefounding has been practiced in the western world. The first Latin types produced by William Caslon in England around 270 years ago were made the way virtually all movable types had been made up to that time: they were cast one character at a time, each by hand. But typefounding has never been just a simple matter of molding hot metal. In fact, the production process that was used in Caslon’s time was painstakingly intricate and included no fewer than four distinct tasks, each involving a separate set of skills.

The first task, known as punchcutting, was generally recognized as the most difficult of the four, an employment that called for meticulousness, perseverance, and extraordinary technical control. It took the longest to learn and it involved the highest degree of artistic scrutiny. The work was tedious and called for steadfast precision in addition to acute eyesight. Because of its exacting nature, punchcutting was ordinarily performed to best effect under a magnifying lens.

In essence, the job required a proficient engraver, like Caslon, to carve in relief the mirror image of a letter, figure, or other character of the desired type size on one end of a shank of annealed steel, later to be hardened by means of tempering. Ultimately, this hardened shank would be used as a punch for striking a matrix, or a number of matrices, from which to cast that one size and style of character needed for printing. Admiration of the craft notwithstanding, the making of punches was, in itself, strictly a means to an end. Its objective was that of achieving (albeit far less directly than one might wish) a three-dimensional impression of a character from which type could finally be cast.

Invariably, the punchmaking process began not with the immediate carving of a desired character, but rather with the creation of various tools and implements needed to fashion the final punch. For example, engraving instruments (gravers) first had to be made and carefully sharpened for the job. Also, files were needed to efficiently remove quantities of steel surrounding a character and help to give it its outer shape; the process was seldom a matter of using gravers exclusively. Files were utilized especially for delicately refining an array of outward-facing, flat or convex edges, such as the top and bottom of an H or the outside of an O. Furthermore, the whole process typically involved making special preliminary punches, called “counterpunches,” to stamp the hollow, recessed, interiors of characters, those non-printing white spaces like the upper and lower rectangular counter forms inside an H or the ellipse inside an O, which at small sizes were tricky (if not impossible) to gouge out accurately using a graver, and were too confining for a file.

Since the method of letterform sculpture employed in punchcutting was basically subtractive, meaning that soft steel was being taken away from negative areas instead of being added to positive areas, accidentally cutting too deeply could be costly. One little slip of the graver could mean the loss of more than a day’s work. The only way to effectively get back lost material that had been removed by mistake was to file down the entire face of the character uniformly, as far as necessary, and try again. Fortunately, there were cures for some other mishaps, as well. If a punch had been tempered, but subsequently found to require additional cutting, it could be softened again by annealing it. A punch that got broken could occasionally be repaired or recut to form another character, depending on the nature of the break, and so on.

In all, the range of preparatory effort necessary for punchcutting and matrix-making was considerable, and despite the fact that multiple matrices could be struck from a single punch, the harsh reality was that a separate punch had to be created for each character in a font of a given size. Punches of virtually every size had to be engraved, filed, proofed, examined, reworked, and reexamined under the lens in fine detail before they were considered finished and ready to be hardened and used for striking.

The second task in the process of producing type, matrix making, demanded somewhat less artistic skill but still an fair amount of ability, care, and knowhow. Here again, Caslon had an advantage, having been well trained in the art of chasing; a practice of ornamenting metal using a hammer and tools designed to indent, rather than cut, the surface. Since striking a punch to form a matrix could easily cause the punch to break, experience was important and caution was essential. In striking, the tempered punch was driven, rather like a headless spike, into a block of (comparatively softer) copper or brass in such a way that the characteristics of the original cutting would transfer in their entirety and would accurately reflect a well-defined, sunken, right-reading impression. From this stamped recess or cavity, the actual types were formed. Before the matrix could be fit into a type mold, however, its sides and face often had to be filed down flat to return the block to its original rectangular shape (which ordinarily got distorted during striking due to a bulging effect caused by outward displacement of the matrix metal) and to give the strike its desired depth. Also it was essential that a matrix be justified (filed) to the proper dimensions. Not only did it have to correctly fit the mold in which it was to be positioned, it had to be calibrated so that it would match the other matrices in its font.

The third major endeavor in the manufacture of printing types was the actual casting. To start, a matrix was placed into the hand mold and the chamber of the mold was adjusted to give the body of the cast character its exact dimensions, which were necessarily different from those of the matrix. Traditionally, individual types were cast from a molten alloy of lead, tin, and antimony. Practiced workers would pour, mold, and remove one identical piece after another, repeatedly, thereby building up an almost endless supply for printing. Each piece of type made in this way was intended to be a faithful three-dimensional reproduction of the character that had been fashioned on the punch.

A fourth operation, dressing the type, entailed a handful of minor tasks. Preparing type for the press included breaking off the jet and filing off any other protrusions left from the mold, reducing the type to the proper height, and sometimes placing a nick at the front of the body to indicate its type size and true alignment.

Although dressing type demanded minimal skill and aesthetic discretion, there was, before Caslon and since, a need for any individual involved in the production process to acquire and maintain awareness of good technique.


Historical Background of the Original Types

When, in the early 1720s, engraver William Caslon first turned his practiced hand to the art of cutting punches for type, England was not particularly noted for typefounding. Printing, as a trade, was thriving in England, but there was not yet an identifiably English style in types. Rather, the types then in fashion were of a class now called “Dutch Old Face,” which were used for both mercantile and scholarly printing. The majority of these faces dated, at least stylistically, from as early as the late 16th century, with later cuttings being added during the 17th century. So popular were these designs that all of the best printing types available in England at the beginning of the 18th century were being imported from Holland. Caslon’s typefounding venture rapidly changed that; it succeeded in providing a domestic source of superior printing types and signified the emergence of an English style.

To the extent that Dutch typefaces of the late 16th century are known to have been modeled on slightly earlier French faces (designs likely to have been cut by Claude Garamond, Robert Granjon, Simon de Colines, and others), it would seem reasonable to assume that punchcutters in Holland may have continued to look to France for typographic inspiration well into the 17th century. There was, in fact, quite a lot of crossover taking place between the European countries. In Amsterdam, punchcutter Dirck Voskens was reputed to have taught Hungarian visitor, Miklós (Nicolas) Kis, the art of cutting letters for type. Both men contributed independently to Dutch type development, Kis being the creative talent behind some later famous types which were long attributed to Dutch punchcutter Anton Janson and bear the name “Janson.” Original matrices for several sizes of that type survive in Germany.

Neither the accomplishments of Voskens nor Kis, important as they were, overshadowed those of an older Dutchman, Christoffel van Dyck. Van Dyck, the greatest cutter of Old Face types in Holland, active between 1648 and 1670, demonstrated in much of his work an acute appreciation of Parisian types, many then already as old as a hundred years. Nowhere is Van Dyck’s admiration more apparent than in his italic fonts, which showed that he excelled at cutting swash letters. While French models were the sources for Van Dyck’s letters, he branched off into other, stylistically Dutch, areas of experimentation and development, producing an array of faces. Of that diverse group, the roman faces seem to have had the strongest impact on Caslon.

One can observe that some degree of French influence was still present in many Dutch designs at the end of the 1600s, when Old Face neared the culmination of its development in the Low Countries. It would therefore seem plausible that Caslon’s earliest text types, in more than an indirect or incidental way, similarly owe the merits of their precision in cutting and fitting to French styles. Without argument, William Caslon already possessed superior technical abilities when, several years into his working life, he turned his attention to cutting punches, proving himself to be the finest talent in England up to that time. He also must have had some very good examples of printing from France and Holland on which to base the fitting of his faces. Cutting and fitting are two critical points on which he was able to elevate himself above his English predecessors and ultimately establish a clear advantage over his foreign inspirers.


Early Attempts to Imitate Caslon’s Style

Caslon types became prevalent in England and throughout her colonies in the latter half of the 18th century. For the most part, William Caslon had no serious rival. Attempts by other English typefounders to imitate his styles did cause the Master some displeasure, but he seems to have had the good fortune of staying ahead of his domestic competitors, and above the rise of foreign ones. A sizable share of types cast at the Caslon foundry were exported to the Colonies, where major production of counterfeit versions did not yet exist. For a time, the business stayed at home.

In America, Caslon type was in favor during the Colonial period but fell from favor, as it did in Europe, when Transitional and Modern faces became fashionable at the end of the 1700s. Old Face types remained out of favor through the first half of the 19th century.

According to type historian, Alexander Lawson, “...the great resurgence of Caslon type in the United States can be dated from 1858, when the Philadelphia foundry of L. J. Johnson (later MacKellar, Smiths & Jordan) brought fonts from England and duplicated them by manufacturing electrotype matrices, a process by which a founder could duplicate a competitor’s type without cutting punches.” This occurrence did not, alone, signal the demise of the punchcutter but it foretold of changes. Within the span of a generation, typefounding technology advanced again.

In the mid-1880s, the American inventor Linn Boyd Benton developed a pantograph machine that could engrave punches and matrices in any quantity desired. No longer did an imitator of the popular cuts of foundry type need to spend years learning the art of punchcutting. The role of the punchcutter became practically redundant and the overwhelming emphasis on skill in type manufacturing shifted to making large-scale mechanical drawings for brass or copper patterns of typographical characters that were several times the size of the type to be cut. Once the drawing of a letter had been finished, and its metal pattern engraved, that pattern could be traced by a pantograph operator to make unlimited multiples of the character in a range of sizes, not just one.

Compared to the exactitude and high art of cutting punches, drawing large letterform outlines for patterns to be produced on the pantograph was a far less demanding and more forgiving enterprise. Quality was certainly a consideration, but slight flaws or irregularities in the contour of a pattern ordinarily did not spoil the pantographic interpretation at text and small display sizes. Hence, minor imperfections hardly mattered, as they had in cutting punches by hand. For this reason, a person less expert than a punchcutter, a person perhaps not as sensitive to the myriad, infinitesimal, sculptural details of three-dimensional type, but competent with respect to the mechanical drawing or cutting of an essentially two-dimensional representation in enlargement, could be readily hired and assigned to this new occupation. The business of typefounding, which had for over 250 years required the talents of an able engraver, eventually grew to attract mechanically-minded lettering draftsmen, first of a breed of 20th-century type designers.

Inevitably, the production of imitation Caslon types by means of the pantograph engraving machine proliferated. Manufacture and export of authentic Caslon fonts, once the exclusive domain of the Caslon typefoundry, ultimately declined, many of the original fonts having become the prey of any pattern maker who wished to make letterforms in their likeness. Numerous letter foundries in the U.S. got in on the act and, as the years passed, nearly all type companies in America offered their own versions of Caslon.


Caslon Revivals

Caslon revivals in America and Britain have enjoyed remarkable sustained success. Throughout type’s recent history, few designs have been as often selected by typographers and printers for use in both text and display. A few of the most popular 20th-century versions include the American Type Founders ’Caslon No. 471, Monotype Caslon series 20, and ATF Caslon No. 540. All have points worth mentioning.

Caslon Oldstyle No. 471 is the metal version considered closest in lineage to William Caslon’s originals, due to the fact that No. 471 was based on the L. J. Johnson holdings mentioned above. The ATF Specimen and Catalog of 1923 devoted a full page to the connection between their casting and the original, emphasizing the accuracy and authenticity of their own. It should be noted that ATF, as the second successor to the Johnson Type Foundry, claimed Johnson had imported original matrices, not that he had made counterfeit matrices from type he imported, as was later asserted by Alexander Lawson, previously quoted. Below is the text of the ATF specimen page in its entirety.

“The American Type Founders Company makes Caslon Oldstyle Romans and Italics precisely as Mr. Caslon left them in 1766, casting the letters from the original matrices, including all the ancient quaint double and long letters and ligatures used during the lifetime of Mr. Caslon. As is well known to many who read of such matters, these types disappeared from the English Specimen Books at about the year 1800, and did not reappear until 1859, in which year the matrices were brought to America and used by The Johnson Type Foundry, afterwards MacKellar, Smiths & Jordan, of Philadelphia, the immediate predecessors of the American Type Founders Company.”

Whether or not ATF’s Caslon Oldstyle No. 471 has the distinction its manufacturer claimed is a matter for debate; but No. 471 did serve a need. In England, where there was also a market for Caslon, an imitation of the Johnson version was also produced by the Monotype Corporation: their Caslon series 20, issued in 1903. It was, according to American typographer and printing authority Bruce Rogers, writing in 1917, “a very close approximation of Caslon Old Face.” But as another writer has pointed out, Monotype’s series 20 had “the disadvantage that all the founts were repetitions of a single original size.”

This was also the opinion of Rogers, in retrospect, who remarked in his book, Paragraphs on Printing, published in 1943, “It is a truism that almost every face of type has its ideal size, and lessens in merit as the size is either increased or decreased. The modern practice of cutting all sizes (at least down to 8-point) from one pattern on a pantograph machine is accountable for much of the mechanical appearance of our books. But before the pantograph was invented, each size, although based on one model, was really a separate design. For the punch-cutter was then in effect the ultimate designer of each size he cut, in proportion to the fidelity with which he followed his model. Some founts matched up very badly, but most of them (Caslon for instance), though the definite family likeness ran throughout the sizes, varied just enough to avoid a sense of monotony.”

This problem has persisted and is present in many typeface revivals today, mainly those with one master for a series of needed sizes, a practice which at best does justice to only one of them, or short of that, fails completely.

A second effort by ATF to produce a Caslon face, and in the process improve on No. 471 led to Caslon No. 540. It was perhaps the most successful and most widely imitated of the ATF Caslons, a version that was made to be more regular in appearance than No. 471. Released in 1906, three years after Monotype Caslon series 20, Caslon No. 540 was ATF’s answer to a plea from advertising printers for a face that would set more compactly. Caslon No. 471 was used as the starting point, and what resulted from the makeover was a new face with more consistent color, shorter descenders, lining figures substituted for old style figures in the primary font, and noticeable changes in the features of a very few characters. Of all the variations of Caslon types introduced in this century, Caslon No. 540 survives in more places and has existed on more typesetting systems than any other.


Wood Caslon

These versions were few in number, but not unknown. In his book, American Wood Type, author Rob Roy Kelly wrote of the Hamilton Manufacturing Company’s mildly-successful attempt, “This design was cut by Hamilton and put on the market after 1900. It appears to be patterned after metal types of the founder. Lowercase is missing.” Missing also seems to be Caslon’s sense of structure in a few of the letters (R,S,X), and most of the figures (2-9).


Caslon for Lettering

Caslon revivals inspired three prominent 20th century American lettering artists, all of whom also wrote books about their craft. Each of the letterers exercised artistic license, to one degree or another, and offered his own Caslon take-off for followers to imitate. Represented are interpretations from three periods of the 20th century.

Lettering for Commercial Purposes was written and extensively illustrated with original hand-lettering by one of America’s most capable showcard writers, active in the early decades of the 1900s. While the lettering styles of the author, William Hugh Gordon, were fairly typical of the period, Gordon himself was not. He became enormously influential. Not only did he make his mark on the lettering arts through his own work, but also in collaboration with another, younger, lettering artist named Ross George, later the principal author of the Speedball Textbook (a veritable primer for two generations of lettering practitioners).

Traditionally, those who make a living through their lettering have looked to different sources for inspiration. Caslon styles of type, widely available across the U.S. in the early 1900s, were among Gordon’s favorite typographic models. His hand-lettered rendition of Caslon Old Style, and a rather more loosely interpreted Caslon Old Style Italic, were offered to his readers for suggestion. If they had been intended merely as copies of Caslon and had been judged only for their fidelity to the model typefaces, these published plates would surely have been targets of criticism, but in the context of showcard writing, a profession that placed a premium on speed and grace with a brush, the Gordon examples served well. The examples also make an interesting study in graphic expression.

Tommy Thompson, one of the best known and certainly one of the most competent lettering artists of the middle years of the 20th century was also typographically knowledgeable. Not surprisingly, Caslon was his choice of type as a basis for formal roman lettering. “Caslon type is herein used as the model from which to study as it remains the type style used to print the greatest body of literature in the English language,” he wrote in the mid-1940s. Like Gordon a generation before, Thompson published his own hand-lettered interpretation of the Caslon style in his book, How to Render Roman Letter Forms, but because Thompson was in the business of creating headlines for photographic reproduction, he was far more practiced at meticulously rendering letters in pen and ink. His method was typical for most professionals of his day.

A few of several contemporary and popular Mike Stevens styles of brush lettering for signs shown in his book, Mastering Layout, are based on typefaces, and not just a single face. Of one such style Stevens wrote: “This alphabet was inspired by Caslon Italic, but in my own work has gradually evolved in appearance towards Century Italic. It may be used as a display or a text alphabet. The thin strokes are made with the contracted tip of the brush. With a little luck and some touch-up, the bold stroke can be rendered with a single (brush)stroke...” 1


Film Caslons

Phototype imitations of 20th-century metal interpretations of Caslon types were the most common versions available when filmsetting was the dominant technology, a decade or more ago. During the era of phototype, the only major manufacturer of typesetting equipment to offer a fresh interpretation of Caslon to the typesetting trade was H. Berthold AG of Berlin. Practically all other manufacturers offered only designs that were copied from faces that had already existed. However, large typesetting services like Photo-Lettering, Inc. (which for most of their years in business did not sell their fonts), introduced dozens of new Caslon variants and Caslon hybrids for the use of their customers. For Photo-Lettering and others who designed their own typefaces for filmsetters, it was a time of great experimentation in style.

One so-called Caslon represented in many phototype libraries was Lubalin/Smith/Carnese’s ITC/LSC Caslon 223, an exaggerated, highly-animated, nearly descenderless display face. Jazz drummer and typeface designer extraordinaire, Ed Benguiat, of Photo-Lettering, was commissioned by International Typeface Corporation to draw a series in a related style that could be used for text in addition to display. ITC Caslon 224, released in 1982 for various manufacturers to adapt to their individual phototypesetting and digital typesetting products, was the eventual outcome. It reflects more of Benguiat’s bee-bop buoyancy than Caslon’s careful control. Call it Jazz Caz.


Digital Caslons Reviewed

Big Caslon is the name of Matthew Carter’s digital characterization of the Caslon firm’s mid-18th-century specimen showings of large roman display types, which because of their lack of strict stylistic conformity with William Caslon’s smaller roman faces, may have been the work of other hands. Indeed, the lowercase alphabet that was paired with the 48-point (or equivalent) capitals and shown in more than one of Caslon’s broadside specimens is known to have been cut in the previous century by Joseph Moxon, his Great Canon of 1669. In the Big Caslon specimen issued by Carter & Cone, this uncertainty is addressed, allowing that the digital version is based in part on forms of unproven origin. Big Caslon was released in one weight, without an italic.

Adobe Caslon represents an astounding attempt to sum up a range of William Caslon’s text faces in a single digital master, and to dream up some fanciful additions in the Caslon style where Caslon himself had evidently been inclined to leave well enough alone. It was, on the whole, an ambitious project by designer Carol Twombly, who cannot be faulted for many of the compromises.

Yet, having had the technical means to produce a digital series of each of the five sizes studied for the purpose, and even going to the trouble of “making careful sketches of every letter at each size” (sketches that could have been used to preserve the integrity of each Caslon original) it is a pity that Adobe instead opted for synthesis and consolidation. In this respect, it is also ironic that so much effort was exerted elsewhere to build up and fill out a grand total of 21 separate Adobe Caslon character sets, with the result that not one of the sets comprising what pretends to be a “careful revival” is true to any particular font by Caslon.2

Not Caslon, without question, is exactly what its name implies. It is surely one of the most comical and exuberant works of alphabet art that owes any (in this case, literally every) part of its being to swashy, 20th-century American, Caslon-inspired italics. It seems most unlikely that this collection will be able to float alone (it’s already listing hard to starboard), but it may find a place next to numerous time-tested typefaces, if for no reason other than its outstanding entertainment value. Several interesting touches, whether included by accident or design, make for great amusement: wrong font letters like a monoline L and x, a bold weight F, a “shoelace” X, miscellaneous lower case letters throughout, and even a flopped roman A (with modifications). What 18th century printer’s job case would be complete without a share of unexpected typographic variety?

This collection of initials, now in font format, was created by illustrator Mark Andresen, who rubbed down bits and pieces of dry transfer lettering: flakes, nicks, and all. It was digitized by Zuzana Licko for Emigre Fonts.





1. Mike Stevens’ styles are appearing in more and more places, due partly to the fact that many of them have been copied, digitized, and marketed (not always under license) as fonts for computerized signmaking systems.

2. Adobe contends in their specimen that the actual differences in proportion from one size to another, among the five examined in Caslon’s specimens of 1738 and 1786, were “very slight–” their conclusion being based on Twombly’s study of two relationships: x-height to cap height, and relative lengths of ascenders and descenders. A couple things Abobe neglects to point out may be worth considering.

  • 1. The five sizes cut by Caslon were not necessarily meant to be merely scaled versions of each other; they purposely differ in countless meaningful ways, exclusive of size. What Adobe sees as “idiosyncratic” differences resulting from Caslon’s “handwork” were not simply caused by a slip of the graver or by oversight. Such variations have historically been the fortunate result of a punchcutter knowing how to correct the details of a letter’s shape at each given size to make it read best. This, Adobe has since apparently decided to quietly concede, now that touting “optical size” as a possible design axis has become a cornerstone for promoting their Multiple Masters font technology.

  • 2. It can be easily demonstrated that specific characteristics of letterforms can and do differ enormously from one font to another even if the ratio of x-height to cap-height is similar and the ascender and descender lengths remain relatively consistent. For Adobe to draw a broad conclusion about individual letter proportions in five demonstrably different fonts cut by Caslon, based entirely on the two (together inconclusive) comparisons noted in their Adobe Caslon specimen, is nonsense.