Part II

At the Oxford University Press

From about 1864, type matrices were made in the Stereotype department of the Oxford University Press, but the process apparently had been discontinued by the early 1900's. According to Don Turner, the present Typefounder at the Press, records there indicate that electro matrices were made for the Press by two firms: R. P. Bannerman and Son, and the Williams Engineering Company(1)--both of London. Incidentally, these firms made the two pivotal casters now in use in the Typefoundry at OUP.

It appears that Bannerman & Son made matrices for OUP from 1909 to about 1934, and that Williams Engineering was the source from 1934 until sometime around 1950. Presumably the Bannerman firm discontinued matrix making or went out of business, causing the change to Williams Engineering—but this is only a guess on my part. Possibly members of the PHS can shed some light on the history of these two firms.

Williams Engineering apparently discontinued matrix making or went out business around 1950, leaving the Foundry without a source of replacement matrices. It was decided to resume the process at the OUP, but this time responsibility was placed in the Foundry. Don Turner, the present Typefounder recalls: "In about 1953 my predecessor, Mr. L. Bullen, was asked if he would have a try at it (matrix making). He was, like myself, a Typefounder by trade, and had never made a matrix before. In the early days when he was trying to grow matrices he had some help from our Stereo Department for the solution for the tanks, and the fusible casting mould that we have was made at the O.U.P. in our Engineers' department". Mr. Bullen's experiments were successful , and the art of electrolytic matrix making was recovered at the O.U.P.

Don Turner still practices his craft as Founder and matrix maker at the Press, the current member of a long succession of professional Typefounders. On graduating from South Oxford Boy's School at age 14, Don came to work in the letterpress machine room at OUP. He began in April, 1946, and four years later came to the Foundry as apprentice to Mr. Bullen; his apprenticeship was six years. On Mr. Bullen's retirement in 1963, Don assumed his present position. Thus he was closely involved with the second beginning of matrix making at the O.U.P, and can speak from experience. I am indebted to him for his description of the process as it is now practiced at the O.U.P typefoundry.

Don Turner explains how the process begins: "First obtain a type of the character to be grown. The face of this type should be in really good condition. It may be necessary to have the face 'touched up' in the Stereo Department, and any kerns or overhangs on the type must be filled to avoid copper growing under; this also should be done in the Stereo Department. The type should be cleaned up with ink stripper cleaner, as any dirt will interfere with the electric circuit and cause a weakness in the copper. The type must then be rubbed up close to the face on all sides."

My own experience reinforces what Don says with regard to the perfection of the original type. What on casual inspection seems a perfectly good specimen seems to develop all sorts of scratches, dents and banged serifs in the process of electrodeposition! And the process is indeed precise. In their description of electroforming, Pinkerton and Carlin say, "The reproduction of fine detail is so good that it can be matched by no other mass production method. For instance, in a microgroove phonograph record, the lateral excursion of the recording needle cannot exceed about half a thousandth of an inch, while that of the smallest modulation it is considered necessary to reproduce for high fidelity is about half a millionth of an inch. Yet the Master can be faithfully reproduced by electroforming, even to the third generation."(1) So the old saying "what you see is what you get" goes here - in spades!

After properly cleaning and preparing the type, it is placed in the fusible casting mould, and molten type metal poured in to form a base around the type. The matrix is grown on this base.

The fusible casting mould is one of those beautifully simple devices that does its job with an elegant minimum of apparatus. Its function is based on certain features of the hand mould and its relation to the matrix, so all three need to be examined together.

The hand type mould as it is known today comes to us substantially unchanged from the earliest know moulds. Moxon's description - in 1683 - matches quite closely moulds in use in the 1800's, and Mike Parker(1) describes a mould in the Plantin Moretus museum dating from the 1580's that differs little from Moxon's.

Basically the mould (fig. 1) is formed from two stepped bars which slide together to form a rectangular chamber, adjustable in width (set), but fixed in depth (body size) and length (height to paper). The face end is closed with the matrix, and the foot end has a slightly restricted aperture (the jet) through which the molten type metal is injected. The jet serves two purposes: to direct the flow of metal directly to the matrix for a sharper face and to form the "break" so that the sprue or jet may be broken off preparatory to finishing the type. The addition of registers and the 'stool' to position the matrix, baseplates for assembling the parts and woods to protect the Founder's hand substantially completes mould.

In early moulds, such as the one Moxon illustrates a male and female gauge assured a parallel sliding fit of the two halves. Later moulds used "ears" or projections above and below the body pieces to register the two halves. 

The essential dimensions of the matrix are the head bearing, or distance from the top edge of the matrix to the top of the type body (the actual measurement is from the base line to the top edge of the matrix) and the side bearing, or distance from each side of the matrix to the sides of the body.

Referring again to the mould, the top edge (head) of the matrix is positioned against the "stool" formed by the bottom plate of the mould and the carriage. This dimension determines the head bearing. Each half of the mould is fitted with a side register which contacts each side of the matrix when the mould is closed (as in this detail) thus the width of the matrix sizes the mould opening, while the location of the letter on the matrix positions face of the type setwise on the body. The width of the matrix then is different for each different set width of character, being made up of two fixed side bearing dimensions plus the width of the particular character.

In the process of driving matrices from steel punches, the punch cutter would obtain copper bar stock for his matrix blanks. Moxon suggests three widths to accomodate the various sizes of letter, made "an n, and for the Thick Letters an m (at least) thicker

than the Patterns were made, for the Steel Punches to be Forged to a size by." The blanks were to be "each about an Inch and a half long, and a Great Primer or Double Pica deep, and for great Bodyed Letters a Two-Lin'd-English deep."(4)

After selecting a suitable blank, the punchcutter drives the punch to approximate depth, positioning the punch centrally on the blank, leaving sufficient head and side bearing to allow for final justification. Early Founders—who often were their own punch cutters—developed a variety of instruments to measure the depth of strike, vertical and horizontal "squareness" of face to body, etc. Moxon gives an extended description of the process, but it is sufficient to say here that the precision of these measurements and the following adjustments determine both the final appearance of the type as printed and the ease of casting by the founder.

Examination of old matrices will show a variety of strategies used by the founder to justify the matrix. "Botching" was the most common. If too much copper were removed from side or head bearings in the justification process, the founder simply took a chisel or punch and raised a "botch", then proceeded to lightly file the protrusion to achieve the correct dimension. Matrices are also seen with their sides filed off at an angle to correct for the punch having been driven slightly askew.

Molds for machine casting paralleled the design of hand moulds, having essentially the same features. Figure 6 shows a typical mould for a Bruce style typecaster. The only essential difference is the lack of baseplates and woods, and the addition of attachment points for the mechanism.

Electrotyped matrices were developed at a time when the Bruce Typecaster was only beginning to compete with hand casting, and consequently duplicated the usual foundry matrix. A number of methods for mounting the type and forming the matrix were used, including schemes for depositing whole alphabets(5). The fusible casting mould was the most effective solution for foundry style matrices.

The Fusible The Fusible Casting Mould (Fig. 7) provides a support for the "master" type, and a chamber around the type into which molten type metal is poured. This forms the base on which the copper is deposited to form the matrix. The base of the fusible casting mould consists of horizontal and vertical body pieces about 3/8" thick, which determine the overall thickness of the casting. The back or foot side plates are also about 3/8" thick to provide support for the type and to fix it squarely in the mould. The face side plates are about 1/16" thick. Both face and foot side plates are adjustable to accommodate the various body sizes (Oxford's has a range of 3 1/2 to 72 points) but are fixed in side and head bearing dimensions to match the type molds in use at the Foundry. Although these dimensions are similar from foundry to foundry, they were never standardized as far as I can determine.

In fact, it seems that each founder made many of his own tools, apparently from having seen or had described to him what other founders were doing, and so there are minor differences in construction of the same tool from foundry to foundry. 

The micrometer depth adjustment is a case in point. Oxfords Fusible Mould is illustrated to the right. The micrometer adjustment provides for the correct "depth of strike", and bears against the face of the type. The screw at the foot of the type serves as a clamp to hold the type against the micrometer. A similar Fusible Mould used by Stephenson-Blake has only one screw adjustment, and it bears against the foot of the type.

The upper unit of the Fusible mould provides for enclosing the type and determines the second side bearing dimension. It too is adjustable for body size to match the base piece. Here we show the two pieces assembled and ready for the cast. Don Turner describes the adjustment and use of the Fusible Mould: "The type must be rubbed up close to the face on all sides. The type is then placed in the fusible casting mould with the micrometer adjustment set to give the required 'depth of strike'. To obtain this 'depth of strike', take the type mould in which the type will be cast and measure both halves of the mould (the measurement is made from the foot to the shoulder of the mould). If the halves differ in size, take the greater. To this dimension add the 'depth of strike' plus an extra four thousandths for removal with a hand file at the justifying stage". The calculation is:

Top half of mould .8700

Bottom half of mould .8840 (the greater)

Oxford University Press height to paper: .9395

Mould Height: .8840

Depth of strike: .0555

Excess for justification: .0040

Micrometer Setting: .0595

Type with kerns or overhangs presents special problems. Here we see a lower case "f" positioned in the mould. Metal can escape in the space to the right of the body. The usual practice is to fit a space tightly under the kern as illustrated here. (click on the illustration for a larger view)

In the case of swash or some Italic characters with kerns on both right and left, spaces must be fitted to both sides of the type. The space back of the kern must be sealed with wax, as any opening exposed to the electrolyte invites the deposition of copper. Thus Don Turner's admonition that "Kerns or overhangs must be filled in to avoid copper growing under."

Obviously this additional space makes the matrix physically wider, having the effect of increasing the side bearing. Don Turner notes that if the overhang is small, the side registers of the mould can be adjusted to obtain the proper set width. However, the best practice is to either remove metal from the fusible base before deposition to deposit a correctly sized matrix, or to dress the matrix to correct width in justification. This minimizes the adjustments required while casting, making the casting of a full font of type much easier for the caster operator.

After properly adjusting the fusible casting mould and mounting the type in it, molten type metal is poured into form the base on which the copper is deposited. Oxford's fusible mould has the equivalent of the jet at its top, giving a fusible with a 'break'. The fusible base is cut off at this point, and the type body cut through flush with the back of the fusible base. The fusible base and type are then given one final inspection, for as Don Turner says "Any defect in the fusible is faithfully reproduced".

The completed fusible base must now be prepared for deposition. A copper wire is soldered to the back of the fusible, and strips of fairly thin but stiff plastic (1/8 or 1/4" Plexiglas is good) are arranged around the four sides as shown in figure 12. The copper will be grown to about 1/8 inch thick, so the plastic should extend from 1/4 to 5/16 inch above the face of the fusible. A wide rubber band is stretched around the assembly to hold the plastic tightly against the fusible base. The 'back' edge of the rubber band is cemented together to prevent copper growing on the back of the fusible, and it is ready for the depositing tank. (click illustration for a larger view)

Oxford uses a rubber lined tank one foot wide, two feet long and one foot, six inches deep. It holds something more than 16 gallons of electrolyte composed of 16 gallons of distilled water, 32 pounds of copper sulfate, 80 fluid ounces sulfuric acid, and 32 ounces potash alum (potassium aluminum sulfate). This is a fairly standard copper electroplating solution, with the alum being added to improve the 'throwing power' of the electrolyte, i.e. to reduce the tendency of the deposit to grow nodules or 'trees' and hence promote a smoother deposit. A comparison of electrolytes from various sources is included in Appendix 3. Fortunately for the private founder, good deposits can be obtained from a fairly wide range of electrolyte compositions.

In fact, at Oxford no continuing tests or controls have been found necessary to assure correct electrolyte composition. After the initial mixing of the solution, the liquid level in the tank is noted and distilled water added as necessary to maintain that level. Solution control is critical in high production electrotyping operations where high current densities and continuous agitation are used, but less necessary with still baths and the low current densities typically used by matrix makers.

Just as sound copper deposits can be obtained over a range of solution compositions, and the same thing can be said for tank voltage. According to Don Turner, at OUP depositing begins with a tank voltage of 1/4 volt, to produce as smooth an initial deposit as possible. After a few days the voltage is increased to 1/2 volt, and held there until the deposit is something over 1/8 inch thick. Total depositing time is typically two weeks, seven days at 1/4 volt and seven more at 1/2 volt.

Some references indicate tank voltages of one to two volts, but these generally refer to electrotyping where fairly thin shells (about .010 inches) are grown at high current densities and with continuous agitation. For long term depositing typical of matrix making, voltages greater than .50 volts promote the growth of nodules or trees at the expense of other parts of the deposit. Thus the matrix may have thin or weak spots that can burn through in casting.

After the deposit is sufficiently thick, it is removed from the tank, separated from the base, and finishing can begin. Don Turner explains: "When it is estimated that the copper is about 1/8" thick it may be removed from the tank. It should be washed under running water to remove all acid solution. The covering should then be removed and some solder run on the back of the grown matrix. This is to strengthen the copper while being separated from the fusible, as it has been found that the copper shell is brittle and may crack."

"The shell then has to be backed up. Two methods are known: One is to place the grown shell in a special holder while molten zinc is poured on the back of it; the solder which is on the back of the shell causes the zinc to adhere. The resulting product then has to be squared and justified. The other method is to square up the shell and rivet a piece of flat copper onto the back of it. This also has to be justified."

Don Turner justifies his matrices by hand, using files, a micrometer, depth gauge, and engineer's square. The most difficult dimension is the depth of strike, which is achieved by rubbing the face of the matrix on a large file. Don allows .004" for final justification, a rather large amount to remove by hand in my experience. Given the accuracy of the fusible casting mould, head and side bearings should be correct unless spacing was added to accommodate kerns. If needed, the side bearing is adjusted by filing the appropriate side(s) of the matrix. As in other precision operations, patience and frequent measurement are the secrets of success.

Now the shell is "backed up" to get the proper thickness of the completed matrix. If a zinc backing is to be poured, the shell is mounted in a fixture and the molten zinc poured in. If a solid backing plate is to be applied, the shell must have its back dressed flat and parallel to the face, again with recourse to file, square and micrometer.

The complete matrix is then squared up, and trial casts made to verify justification. At O.U.P these casts are made on the Pivotal Casters. If the new matrix is a replacement for one of an existing font, it is checked by comparing the newly cast type to letters from the original matrices with an alignment gauge. As an additional precaution, it is proofed in a line with its mates (an error in alignment of less than .001" is detectable in the printing). Now, if the work has been done accurately enough, the new matrix takes its place in the safe with its neighbors.

This is the process of electrodeposition matrices as it is now practiced by Don Turner at the Typefoundry at the Oxford University Press. The methods are, of course, adapted to the production of foundry style matrices, but are never the less applicable in principle to other styles of matrix. The fact that Don finishes his matrices entirely by hand shows what can be done by careful working, and proves a procedure that is within the capability of most private typefounders. --After all, the fact that we can successfully operate typecasters shows that we have some technical competence!

In the final part of this paper, I will describe methods that I and other private matrix makers have found effective. These comments should provide additional clarification of the procedures used at Oxford, and give the interested reader enough information to begin depositing matrices on his own.

(1)R. P. Bannerman & Son, Northampton Works, Woodgreen, London Williams Engineering Co. Ltd. Nodis Works, Ealing, London, this firm manufactured the Nodis Typecaster.

(2)H. L. Pinkerton and Frank X. Carlin, "Electroforming" in Electroplating Engineering Handbook, New York, Van Nostrand Reinhold, 1971, ch.17.

(3)Mike Parker, Typefounder's Molds in the Plantin-Moretus Museum, in The Library, Volume XXIX, No. 1, March, 1974. pp. 93-102.

(4)Moxon, p.152.

(5)Filmer, "James Conner" p. .

go to Part 3