The Issue of Patrix Cutting in Soft Metal

A Survey of the Errors

image link-topic-sf0.jpg

This is a survey of important instances in the telling of the history of type where the role of patrix cutting and matrix electroforming has been either ignored or misrepresented. It is a part of the Notebook on the history of Issue of Patrix Cutting in Soft Metal.

Sadly, this survey is far from complete. It is almost impossible to find a book in English which contains an accurate account of how type was made.

1. Bullen

Of the people who were alive during the emergence of machine type-making technologies in the late 19th century, Henry Lewis Bullen was one of the most prolific writers. Many of the accounts still repeated today find their origin in him. Unfortunately, he is an unreliable source. Much of what he said is demonstrably incorrect, and this casts doubt upon everything that he said.

Here I'll discuss by topic several of the problems in his stories. The one article which has caused the greatest damage to our understanding of the history of type, however, is:

Bullen, Henry Lewis. "Linn Boyd Benton - The Man and His Work." The Inland Printer, Vol. 70, No. 1 (October, 1922): frontis, pp. 60-64.

I'll refer to it here as [ BB 1922] It is worth reading in its entirety. For a more comprehensive study of this article (which in part duplicates the material here), see the discussion of it in the Linn Boyd Benton Notebook.

1.1. No Punchcutters?

In [ BB 19822], Bullen traces Benton's involvement with type-making technologies back to 1882, saying that "In 1882 Benton's thoughts turned toward the invention of a typesetting machine." This would be a fascinating event in the history of the development of composing machines, but there is no record this in any source other than this sentence.

Bullen uses this machine to explain the origins of what the Benton, Waldo type foundry later sold as so-called "self-spacing" type. This type would more accurately be termed "point set" type - types cast to just a few regular widths rather than each to a width determined by the form of its letter. ([Benton's typesetting machine would effect] "a near approach to self-justification ... by casting all the characters and spaces and quads of a complete body type font of roman and italic on nine widths, instead of on the more than one hundred widths found in the average font of body types. This was ... basically the same [idea] as that which Lanston successfully employed in later years on his monotype machine.")

There's a lot being advocated "under the covers" here - it's a long way and a lot of computing machinery to go from from point-set types to automatic line justification - but that is a subject for a different critique. What is important here is that for whatever reason Benton did to on to manufacture and sell "self-spacing" (point set) types for hand setting (not machine composition). They were not as successful in the marketplace as Bullen suggests, but they were made and sold.

From a typefounder's perspective, simply making point set types is easy: for each sort, you just adjust the set width so that the type body contains, set-wise, the full face and beard of its letter. (This, actually, is what the Lanston Monotype composition caster does - with a fine-grained system which divides the Em of each body size into 18 units.) But if you attempt to do this with a limited number of set widths, as Benton did, it is unlikely that your existing letterform designs will fit well. This is a nontrivial design undertaking, because our letters do not necessarily fit well into a limited number of widths.

So whether or not Benton ever contemplated a typesetting machine, Bullen is correct in observing that equipping a type foundry for the production of point set types (which look good) will require making new matrices. Moreover, Bullen's math actually underestimates the job. As I hope to show, 3,000 matrices doesn't even come close to what Benton needed.

1.2. Benton's Extraordinary Problem?

From Benton's need for several thousand matrices, Bullen makes two arguments. What he writes is:

"Punches for every character were required to be engraved and matrices made. There were more than three thousand punches to be cut and not one punch-cutter was available either in America or in Europe. This dilemma was the turning point of Benton's career - it eventually disclosed to himself that he had mechanical genius of the highest order." (61)

I would distill this into two claims:

First, that punchcutting was a rare skill and punchcutters were not to be found.

Second, that the production of so many matrices was a problem of unprecedented magnitude which led Benton necessarily to develop his pantograph(s).

Both of these points are false.

1.2.1. Patrix and Punch Engraving Not Rare

Bullen's account is complicated by his failure to distinguish clearly between patrix-cutting in typemetal for electroforming matrices and punch-cutting in steel for driving matrices. He clearly uses "punchcutting" at times when he means both. But it is clear that both punch-cutting (in steel) and patrix-cutting (in typemetal) were well established professions at the time. The assertion that "not one punch-cutter was available either in America or in Europe" is contradicted at very least by the existence of a flourishing typefounding industry on each continent. If you tally up the American typefoundry histories in Annenberg's Type Foundries of America and their Catalogs, you find that in 1883 there were at least 39 type foundries in operation in the United States alone (not counting, of course, many more in Europe). While some of these were no doubt doing nothing but pirating types by electroforming, most of them were employing patrix or punch cutters.

Writing not long after this, in 1898, William E. Loy, in his series "American Designers and Engravers of Type" in The Inland Printer wrote:

"The active development of the type founding industry has produce a large number of ingenious designers and engravers of type, about whom the public knows very little."

He further explains that "The soft composition metal used for the engraved letters [that is, for patrix engraving for electroforming] permits a freedom of manipulation which would be practically impossible in steel, and the engraving or cutting of the pattern letters is much more rapid."

(Both quotations from The Inland Printer, Vol. 28, No. 1 (February, 1898). Reprinted on p. 31 of the Saxe/Johnston edition of Loy.)

In the following 27 installments in his series, Loy gives biographical sketches of those type engravers for whom he was able to discover information. Most of them were active in 1883. Many worked for multiple type foundries.

So Bullen's assertion that "not one punch-cutter was available either in America or in Europe" doesn't square with what we know of the scale and activity of the typefounding industry at the time.

1.2.2. Benton's Situation Ordinary

Moreover, there was nothing unique in Benton's situation. His was one of about 39 type foundries in America, most of which wished to introduce new types (many of which were in fact introducing new types at a great rate). Is there anything in a need for "more than three thousand punches" which makes Benton's situation unusual?

How many new styles and sizes of type does this figure of "over three thousand" really mean? In metal type, each character in each style in each size requires its own matrix. So, for example, if there are 82 characters in a standard font of Lanston Monotype Cheltenham Bold 14 point (there are), that takes 82 matrices. If you want 18 point, that's another 82 matrices. If you want 14 point italic that's another 78 mats (not every size or variant will have the same number of sorts). In 1883, making each matrix required first making either a patrix or a punch. To figure the overall matrix count, you need to know (a) how many type styles (faces and their variations), (b) how many sizes each style is cut in, and (c) how many sorts per size. Multiply these together, and you have the number of matrices a foundry would need to produce these types.

I do not have available to me a specimen of the Benton, Waldo & Co. "Self-Spacing" types, so I must do some approximating and guessing here. In the 20th century, a standard set of matrices as supplied might run to approximately 80 characters. (For example, Lanston Monotype No. 86, the ubiquitous Cheltenham Bold, had 82 characters for machine composition from 5 to 12 point, 78 characters for machine composition at 14 and 18 point, 82 characters for hand setting 14 to 36 point, 77 at 42 point, 81 and 48 point, 77 and 60 and 72 point, and only 46 at 72 point.) At the other extreme, in an article on "Self-Spacing Type" in The Inland Printer (which reads like Benton, Waldo advertising copy), a claim is made that "in a complete font of the old kind of body type there are about 190 widths of bodies." An inspection of the table (in this article) of character distribution over the "self-spacing" set widths suggests that Benton was providing uppercase Roman (26 sorts), lowercase Roman (26), small caps (26), figures (10), and 49 points and other characters, for a total of 189. So let's approximate, and estimate that a standard font of uppercase, lowercase, figures, points, and a few ligatures and other sorts would be, say, 100 characters.

The next problem is to determine how many sizes each face was cut in. Again, a Benton, Waldo & Co. specimen would be best here, but failing that there is a substitute. In a table supplied by the Benton, Waldo foundry for use in Caspar's Directory of the American Book News and Stationery Trade [ Caspar 1889] (to illustrate "Type Sizes"), ten sizes are shown (Double Small Pica, Great Primer, Pica, Smal Pica, Long Primer, Bourgeois, Brevier, Minion, Nonpareil, and Agate). So we know that at least one face of Benton, Waldo "Self-Spacing" type was cut in ten sizes. These present a pretty good range, in any case.

So: 3000 matrices / 100 characters per font gives 30 matrix fonts. 30 matrix fonts / 10 sizes per face gives 3 typefaces.


Bullen's figure of "three thousand" matrices represents the work involved in making and introducing three new typefaces. Yes, it's a lot of work. Typefounding is hard work, and always has been. But that's the point: this is no more work than that faced by any new typefounding enterprise since the 15th century.

It is interesting to note that in the same December 1886 number of The Inland Printer there were showings by two typefoundries of six faces in a total of 25 sizes.

2. Anthoensen (1939)

"The steel punch engraved by hand was the vital feature in all typefounding until the engraving machine was invented." {Anthoensen 1939}, p. iv.

3. Eckman (1965)

Dr. James Eckman was one of the great amateur (in the finest sense of the word) scholars of type of the late 20th century to whom we owe so much. Yet at times he relied upon unreliable sources. Since his own works are so good, and now so often cited as if they were primary sources, it is best to identify those few cases where he fell into error.

In his popular volume The Heritage of the Printer, Volum 1 [Volume 2 is an independent work] (Philadelphia, PA: North American Publishing Company, 1965), Eckman is trapped by two problems: First, by 1965 in America the understanding that patrix cutting and matrix electroforming were common methods of type production has vanished (even though Lanston Monotype electroformed matrices were still in common use!) Second, he relies upon a story fabricated by Bullen. In his chapter on "The Gifted Bentons, Father and Son," he writes of L. B. Benton:

"The machine which brought Benton fame was his punch-cutting machine, patented in 1885, which, oddly enough, Benton did not use for the cutting of punches. He used it to cut type metal, for the Benton & Waldo foundry at this time was making matrices by the electrotype process. But the Mergenthaler Linotype Company asked him to make a trial cutting of steel punches with his machine. Benton did so with success, and the Benton machine was then leased to the Mergenthaler firm and subsequently to the Lanston Monotype Machine Company."

This is almost right, and but for Bullen it would have been right. Benton's first pantograph was indeed patented in 1885. It was leased to both the predecessor of the Mergenthaler Linotype Company and to the predecessor of the Lanston Monotype Machine Company (neither of these companies had settled on their final name at this time, but that is not significant).

But Eckman's surprise that a pantograph might be used to cut typemetal patrices is unwarranted - it was not an "odd" practice at all. Moreover, the story that the Linotype company prompted Benton to try cutting punches in steel is false. Bullen made it up. See the discussion of this in the Notebook on Benton's Vertical Pantograph for Patrices and Punches for an analysis of this fabrication and the Notebook "Bullen on Mergenthaler (1922)" for a more general discussion.

4. Baines & Haslam (2002)

Type & Typography by Phil Baines and Andrew Haslam (NY: Watson-Guptill, 2002) contains a reasonably good general overview of "mechanical punchcutting and composition of metal type" (pp. 76-79). But, still, their account is flawed by what is ultimately a reliance on Bullen.

"A machine for casting types was developed early in the nineteenth century, but the long-term success of the idea was dependent upon the supply of vast quantities of matrices from which to cast type. For that, more punches than could possibly be cut by hand were needed. This requirement was met in 1885 with the arrival of the Benton Automatic Punchcutter, invented by Linn Boyd Benton and R V Waldo of Milwaukee. ... Mechanical composition of type quickly followed Benton and Waldo's invention."

There are several confusions at work here, the most important of which is the conflation of typefounding with hot metal composition.

The "pivotal" typecasting machine was indeed invented by David Bruce, Jr. in the late 1830s and early 1840s. But it wasn't dependent upon anything else for its long-term success - it defined the 19th century typefounding industry and remained at the core of 20th century American typefounding until the demise of ATF in 1993. For the forty years from the mid 1840s to the mid 1880s, the needs of machine typefounding were met perfectly well by hand punch and patrix engraving.

Neither was mechanical composition dependent upon the "supply of vast quantities" of matrices. There were several "cold metal" typesetting machines in use in the 19th century. They were not as successful as the Linotype, Monotype, and other "hot metal" machines, but neither were they complete failures. (Discussions of these machines form an interesting part of Walker Rumble's book The Swifts: Printers in the Age of Typesetting Races (Charlottesville, VA: University of Virginia Press, 2003.))

The Linotype (especially) and the later Monotype (to a lesser degree) did require the large-scale production of matrices. Moreover, punching became the preferred method for manufacturing these - not so much because of the speed of the process, but because engraved matrices were not regular enough between production runs and electroformed matrices were not durable enough in composition sizes. The Benton pantograph was indeed leased to both the Linotype and Monotype companies, but it is not true that they were consequences of pantographic punchcutting.

Ottmar Mergenthaler and Tolbert Lanston simply assumed that they would discover or create an appropriate punchcutting technology. They did not first see a punchcutting pantograph and decide therefore to build hot metal composing machines. In particular, the 1886 "Blower" Linotype went into production using electroformed matrices (from conventionally cast types - to match the type in use at the Tribune). These proved insufficiently durable, so after six months production continued using matrices struck from hand-cut punches. Ottmar Mergenthaler was in the process of developing his own punchcutting equipment (of which no technical details survive) in 1888 when the Linotype syndicate (with whom his relations were strained) discovered Benton's pantograph (which had been cutting punches in steel since at least 1884) and began contracting punchcutting from Benton, Waldo & Co. (They leased a machine in 1889.) For further information on this see Carl Schlesinger's edition of Mergenthaler's autobiography, The Biography of Ottmar Mergenthaler, Inventor of the Linotype (New Castle, DE: Oak Knoll Books, 1989).

5. Consuegra (2004)

The account in David Consuegra's American Type Designs & Designers (NY: Allworth Press, 2002), is rather confused.

[TO DO: Summarize and correct.]

6. Bibliography

Anon. [but certainly Benton, Waldo & Co., and possibly Benton himself]. "Self-Spacing Type." The Inland Printer. Vol. 4, No. 3 (December, 1886): 180.

[click image to view larger]

image link-to-inland-printer-v04n03-1886-12-uw-sf0.jpg

Scanned by DMM from the original.

{Anthoensen 1939} Moxon, Joseph and Fred Anthoensen. The Art of Cutting, Casting, and Preparing of Letter for Printing, with a Neat Representation of a Letter-founder's Work-house, Together with a Note on Typefounding by Fred Anthoensen. Portland, Maine: The Southworth-Anthoensen Press, 1939.

This is a reprint of the 1750 publication in the Universal Magazine (q.v. for a digital copy of it) of an extract and condensation of Joseph Moxon's description of punchcutting and hand typecasting. Anthoensen was a compositor and a commercial and fine press printer with an interest in the history of type (he wrote another book on John Bell's type). Here he contributed a fine introductory "Note on Typefounding."

Caspar, C. N. Caspar's Directory of the American Book, News and Stationery Trade Milwaukee: C. N. Caspar's Book Emporium, 1889. p. 1377.

Digitized by Google from the University of Michigan copy.

Loy, William E. "Designers & Engravers of Type," a series of 28 articles in The Inland Printer which appeared monthly from Vol. 20, No. 5 (February, 1898) through Vol. 25, No. 3 (June, 1900) [but no installment in Vol. 25, No. 2 (May, 1900)].

I have reprinted some of these online. However, any serious student of type needs to have the edition of Loy's articles edited by Stephen O. Saxe and Alastair M. Johnston, published under the title Nineteenth-Century American Designers and Engravers of Type (New Castle, DE: Oak Knoll Press, 2009).