MATRIX MAKING AT THE OXFORD UNIVERSITY PRESS

Part 1 - Historical

From the time of Gutenberg's invention in about 1450 until the mid nineteenth century, virtually all type was cast in matrices that were struck from steel punches. The basic process, which has been described many times, is to cut the reversed image of the letter on the end of a steel bar, using files, gravers, and counter punches. Thus a "die" is formed in the exact shape and relief of the desired type. Proofs of the letter are taken during the process by coating the end of the punch with soot in a candle flame, and impressing it on smooth paper. After the shape of the punch is satisfactory, it is hardened and tempered so that it may be driven or pressed into a copper bar, which then becomes the matrix from which the characters can be cast for the printer.

Needless to say, the punch cutter was such a highly skilled craftsman that his ranks were always sparsely populated. In his description of typefounding in the United States, David Bruce writes, "Since the starting of Mr. White in Hartford, Conn. 1805 but five letter cutters had been produced down to the year 1839 in the United States. They range in Seniority as follows, Richard Starr, William F. Hill, George B. Lothian, George Bruce and David Bruce, Jr. The scarcity of workmen in this department of typefounding is easily explained. It has always been the policy of The leading founders to purchase the punches of the cutters and by this prudent course hold the keys to the business and thereby control competition". In fact, many of the early typefounders cut their own punches. We don't know whether or not Gutenberg did so, but he was originally a Goldsmith, and so had the requisite skills.

After the punch is made, its continued life is not assured. The driving of the punch into the matrix is attended by several dangers. If the temper is not 'drawn' sufficiently, the punch is brittle and may crack or break when struck. Hard spots in the matrix may break or deform the punch. Moxon cautions that the punch should be good steel "well wrought together" and "without seams of iron" and gives explicit directions on how to prepare the steel before cutting the punch. So the founder needed to have the ability not only to make his original punches, but also to replace a punch from time to time.

But why do punches have to be used over and over? Once struck, the matrix should last almost indefinitely. It is made of a solid bar of copper, and is subject only to the wear caused by the fluid type metal. But the combination of lead, tin and antimony heated to over 600 degrees F. is slightly corrosive, and its effect accumulates over the thousands of letters that are cast in the typical matrix. Also the copper is not necessarily homogenous. It may have pinholes or bubbles in its interior that break through after the mat has been in use for a time. Thus new matrices are occasionally required.

In hand casting the problem is minimized because the pressure involved is slight--even the most vigorous 'throw' of the hand caster can't compare with the force generated by a casting machine. The advent of machine casting with its harder metal compositions and higher temperatures and pressures made matrix wear a real problem and increased the need for replacement matrices.

It is interesting to note in this respect that Oxford's historic Fell types are still cast mostly from the original matrices. Don Turner, Oxford University Press Typefounder comments, "When they (the Fell matrices) were made around 1670 they were for use in hand moulds, but we now cast them on our pivotal casters. We do have a very few electro matrices of some of the original Fell matrices, this being because some of the original matrices were made too small or thin in size to work on pivotal casters. I have to be very careful when casting from them because as you say, they are of such historic significance; but it is for this very reason that we still use them".

Thus a historic problem of the type founder has been to obtain and maintain a supply of matrices from which to cast his type, and the hand cut punch was for many years the only agency for striking those matrices. In fact, if the mechanical punch cutter had not been invented, the deployment of modern typesetting machines such as the Monotype Composition Caster and the Linotype would not have been practical. ...But that is another story.

In the opening days of the nineteenth century electrical effects were just beginning to be investigated scientifically. Those were the days of Volta, Watt, Faraday, Ohm, Gauss, Ampere--men who left their names for the basic electrical units. And they were days when scientific discovery was not the sole province of the laboratory. The materials and techniques were simple by today's standards, and experiments could be carried out, if not with 'simple materials found in every home', at least with materials and equipment that could reasonably be made or purchased by the experimenter.

The discovery of electrical effects is lost in antiquity, but it was Volta's demonstration in 1800 of the Volta or galvanic pile, the first battery, that initiated widespread interest in electrochemistry. Many of the early experiments in this new science were carried out with the Volta Pile, but its practical use was limited by polarization (the insulation of one or both poles by hydrogen generated in the reaction) which limited the current that could be obtained.

In 1836, John F. Daniell, professor of Chemistry at Kings College, London, invented a new battery that avoided these effects. The Daniell cell consists of a zinc anode immersed in a solution of zinc sulfate or dilute sulfuric acid and a copper cathode in a solution of copper sulfate. In the earliest Daniell cells the two liquids were separated by a porous wall, typically plaster of paris (a later design took advantage of the difference in specific gravity of the two solutions to separate them by 'floating' the zinc sulfate solution on the heavier solution of copper sulfate). In operation, the Daniell cell generates an electric current by the solution of zinc into zinc sulfate at the anode, and deposits copper from the copper sulfate solution at the cathode.

In September, 1839, Thomas Spencer read a paper to the Liverpool Polytechnic Society entitled 'On Working in Metal by Voltaic Electricity'. In his paper, Spencer reported his experiments with the Daniell cell, and its use in electro-deposition of copper. He describes his experiments thus:

"In the latter part of September, 1837, I was induced to make some electro-chemical experiments, with single pairs of plates, consisting of small pieces of zinc and equal sized pieces of copper, connected together with wires of the latter metal. It was intended that the action be slow; the fluids in which the metallic electrodes were immersed, were in consequence separated by thin discs of plaster of paris. In one cell thus formed was placed sulfate of copper in solution; in the other, a weak solution of common salt.

I need scarcely add that the copper electrode was placed in the cupreous solution, the other being in that of the salt. I was desirous that no action should take place on the wires by which the electrodes were held together; and to attain this object, I varnished them with sealing-wax varnish; but, in one instance I dropped a portion on the copper electrode that was attached. I thought nothing of this circumstance at the moment, but put the experiment in action.

This operation was conducted in a glass vessel; I had consequently an opportunity of occasionally examining its progress from the exterior. After the lapse of a few days, metallic crystals had covered the copper electrode, which had been spotted with the drops of varnish. I at once saw I had it in my power to guide the metallic deposition in any shape or form I chose, by a corresponding application of varnish or other non-metallic substance".

Spencer goes on to describe his further experiments, including depositing a negative image of a coin, and attempting to produce raised letters and images on copper by engraving through a wax coating. Finally he impressed type into a lead sheet, using a screw press, and deposited copper onto this matrix. He says, "I now had the satisfaction of perceiving that I had by these means obtained a most perfect specimen of ,  stereotyping in copper, (Italics mine) which had only to be mounted on a wooden block to be ready to print from".

Spencer, however, was not the only experimenter to discover electrodeposition. The process was apparently discovered independently by Moritz Jacobi in St. Petersburg, as well as Spencer and C .J. Jordan in England. Jacobi's description, also in 1839, was the first to be published, and there was a lively controversy over who actually had first made the discovery. I tend to agree with William Filmer, the writer who published Spencer's description in "The Printer", who said, "Undoubtedly, all three made the discovery independently, and without the knowledge that any other person had produced the same result. If these persons had not noticed the phenomenon others undoubtedly would have within a short period. How useless, then, to quarrel about the difference of a few days, or months, in the date of its publication".

Interestingly, the article on Electrotyping in the Encyclopedia Britannica names Joseph A. Adams, an American, as an independent discoverer of Electrotyping in addition to the above three. This is probably attributable to T. L. DeVinne, who stated that Adams invented electrotyped woodcuts in his "the Practice of Typography - Plain Printing Types". Indeed Adams was active in the technique of electrotyping wood engravings, but apparently had no connection with the discovery of the electrotyping process itself.

Rollo G. Silver traces the development of electrotyping in America following the announcements of Jacobi and Spencer. The practical aspects of electrotyping were developed rather quickly, and by 1856 there were electrotyping plants in Boston, New York, Chicago and San Francisco.

Concurrently with the development of electrotyping, experiments in the electrodepositing of type matrices were being made. As early as 1840, the "American Repertory" reported: "Within the last week, the matrices used by typefounders have been duplicated with entire success".

The typefounder James Conner also was an early experimenter in the making of electro matrices. William Filmer, writing in "The Printer" in 1859 describes Conner's experiments with making both individual matrices, and matrices of complete alphabets. The implication is that these experiments were early, but no date is given. This and other references in Rollo G. Silver's article indicate that electrolytic matrix making was in fairly general use by 1844, but in 1845 the process was patented by Thomas Starr of Philadelphia in U. S. Patent #4,130. (Appendix I)

Annenberg states that Conner in fact developed the process of electrotyping matrices, and that Edwin Starr was involved with the process. Thomas Starr was the son of Edwin Starr, brother of Richard Starr, who Bruce listed as one of the five punch cutters in the United States in the early 1800's. Silver traces the location of the Starrs, father and son, but can make no definite connection between the younger Starr and Conner. It appears probable that Edwin Starr worked in the Conner Foundry, and very likely was involved with the development of electrotyped matrices. Thus it is possible that Conner's idea was patented by Edwin Starr in his son's name. Probably the complete story will remain a mystery.

There is an interesting story associated with Edwin Starr: it appears that he might have been an early industrial spy, although there is no firm evidence of the fact. In the early 1800's Starr was employed by Elihu White and William Wing who were in business to manufacture a typecasting machine of Wing's invention (U.S. Patent dated August 28, 1805). The machine was to have cast complete alphabets, which then were to be cut apart to form the individual types. White and Wing appear to have invested considerable time and money in the machine before concluding that it could not be made a commercial success. Now, having invested in the typefounding field, White and Wing apparently decided that if they couldn't make the machine they might try making type by hand. David Bruce relates a rather amazing conference: "On summoning an advisory council in relation to future proceedings.(sic.) It was found that neither of the partners or any of their workmen could say they had ever seen an ordinary hand mold or had witnessed a single type cast!" Bruce goes on to say that White later learned the secrets of the mold in Scotland, but Annenberg tells a story of industrial intrigue.

At the time, the only operating Typefoundry in America was that of Archibald Binny and James Ronaldson in Philadelphia. They quite obviously were not anxious to share the secrets of typefounding with a prospective competitor. Bruce put it succinctly, "Messrs Binny and Ronaldson who prudently excluded all visitors - having an aversion to these Hartford Yankees". Starr therefore posed as a disgruntled employee of White and Wing, offering to give the secrets of machine typecasting to Binny and Ronaldson in return for employment. According to the story, Starr kept his end of the bargain, giving the details of the unsuccessful Wing machine to Binny and Ronaldson. He also copied the hand mould for his real employers who then became successful Typefounders. And so industrial espionage was with us, even in the "good old days"!

In any case, electrotyped matrices were here to stay - for better or worse. Most established founders damned them - curiously enough the Conner Foundry among them - but it also is probably safe to assume that most used them. The basic problem was one of a good process put to a bad use.

Prior to the advent of electrotyped matrices, the scarcity of competent punchcutters limited the number of founders. Electrotyped matrices removed that limit. Now anyone wishing to could duplicate a founder's matrices, and numerous foundries sprung up whose only stock in trade were other founders' designs.

But the process had its brighter side. In a retrospective look, Carl Shraubstadter, Jr. a designer of type and son of one of the founders of the Central Type Foundry in St. Louis observed: "Though there are many imperfect matrices of this kind, and the comparative ease with which faces can be copied, has tempted a few to open foundries without proper tools or appliances, there is no doubt as to their producing, when well made, as good type as that cast from a copper strike". He goes on to say that the electro matrix is in fact better for type larger than 36 point, indicating that driven matrices in these sizes tend to be hollow-faced; and he lauds the electrolytic process for allowing the production of "the handsome modern faces, with their delicate lines and shadings". The Victorian type styles having gone out of use years ago, we might argue with this last observation!

But Schraubstadter's assessment was correct. The electro matrix was—and is—fully equal to the driven matrix. Virtually every modern foundry utilizes the method to provide replacement matrices, and to "recover" types for which matrices have been lost or destroyed. And it is in this last use that the private founder will find its greatest value.

As the use of hot metal type declined over the past twenty years numberless matrices have been lost or junked. Most

American Monotype display matrices were brass, and to the misfortune of the private founder, they were worth more to a printer closing a hot metal operation as scrap than as matrices. An experience I had illustrates the problem: While on a business trip in St. Louis, I contacted a printer who I had heard had a supply of Montotype composition matrices. And indeed he did have—stacked almost four feet deep between sheets of heavy cardboard on a pallet! He had a standing offer of eighty cents a pound from a junk dealer, and would sell to someone like me in small lots at ninety cents. I immediately thought of buying the whole stack, and counted a few layers to estimate the total quantity. Weighing 10 matrix cases and multiplying, I found he had over $900 in scrap value! I selected what I could carry and pay for, and went on my way.

But more to the point is the fact that not only are the Monotype and Thompson display matrices disappearing, but also many foundries have closed, and those surviving have restricted their operations, making many historic typefaces and ornaments unavailable. Thus electrolytic matrix making by private founders promises to be the only practical method of making these types available to the letterpress printer.

I say private founders because I believe that with limited demand, commercial casting of unique historical faces will be largely impossible - the cost of casting small volumes is just too great. Thus the importance of the private founder, who can invest many hours for the joy of accomplishment, and not expect to be paid adequately for his time.

Part II of this paper describes the process of electrolytic matrix making as it presently is done at the Oxford University Press. The operation is not unlike that of a private foundry in that matrices are made there for private use, and for relatively low volume casting. The greatest part of the work of the Press is, like that of any other major printing house, primarily offset using modern computer typesetting. Hot metal is seeing less and less use, and hand set types are used only for those special publications where time and cost are no object. Thus the methods in use at Oxford's typefoundry are suitable for the private founder working on a small scale.

Part III adds details of the matrix making process as I practice it at the Recalcitrant Press, and the appendices furnish further details that should help others apply the process. Few of the techniques are unique - I have been helped by many generous friends. If the methods described are useful - historically or practically - I will have been well rewarded.

Footnotes for part I