Relabeling Those Pesky Caslon Matrices

I recently found a pair of mis-labeled matrices in my set of 18-point Caslon Old Style, and although I could just leave a note in the storage box, I thought it would be better to just correct the markings on the matrices themselves.

Zero on the left, marked as the letter ‘o’, should have the ‘F’ mark and width *8 8, and the letter ‘o’ on the right, marked as the figure zero, should have no ‘F’ and width *8 2.

I have a set of 2mm (about 0.080″) alphanumeric punches, which I actually purchased for just this sort of work, and this would be my first use of them for this purpose.

One question was how I would obliterate the incorrect markings. The fancy way would be to mill off the marked edge, making the matrix resemble the stamped aluminum style of matrix, and put on a complete fresh set of markings. In this case there were only a couple of individual letters and digits to remove, and plenty of space nearby for the new markings, so I just opted to essentially scribble over them by punching them with an ‘X’ a few times.

Punching the markings raises the surrounding metal a bit and it it essential that any such raised area that is too near the letter itself be smoothed off so it does not impede proper seating of the matrix on the face of the mould. On these particular matrices I have enough room to keep the markings well away from this critical area, but I removed the raised parts anyway so the matrix would have a flat face overall. This also makes the markings easier to read, and it looks like I got the ‘8’ inverted. These punches are pretty small, so perhaps I’ll see if there are any orientation marks on their shank to avoid this in the future.

49A Matrix Holder Parts Breakdown

The Monotype 49A matrix holder is listed with plate 35 in the English Spare Parts List book, and although a list of all its parts is included, the holder is not shown in the illustration. This post shows the internals of this holder and identifies all the individual parts.

Rear view, fully assembled

Note that the adjustment for the mould size has a sort of vernier scale, where you match lines to get the desired position. This allows the individual labels to be far enough apart to be legible. The ‘D’ and ‘E’ suffixes on the sizes would appear to refer to Didot and English point systems. I’m not sure what the ‘A’ refers to, perhaps it refers to an American (Lanston) display mould, which uses the same matrix position in all cases and is the same as the English 36-point position.

Front view, fully assembled, with matrix installed

Rear view, assembled but with cover off

Disassembled with individual parts identified

The parts are identified by part code, but if they were marked with a part number as well this is shown in parentheses. I’m not actually sure where that spring labeled only ‘?’ came from, as there are no points to anchor either end of it to. Perhaps it was just sitting loose inside this assembly when I first opened it up.

Improved Shim for my 49A Display Matrix Holder

When I cast display type from Lanston display matrices I use an English mould because the one I have is in generally better condition than the American-made Lanston display moulds I have.

This mould’s cavity is not in the same position (except for 12- and 36-point casting) as that of the Lanston moulds, so English Monotype has a special matrix holder, designated number 49A, for this purpose. The holder has an adjustable slide on its back, which one sets to the position appropriate for the body size of the type to offset the matrix position by the correct amount. Each time this is adjusted one also has to adjust the front draw rod on the caster so that that cone hole on the matrix holder gets rough-positioned under the centering pin.

In addition to the difference in mould cavity position, there is also a difference in the height of the face of the mould above the caster table, or perhaps this matrix holder holds the matrix a bit high. In either case, if you have the carrying frame on your caster adjusted for using Lanston matrices, moulds, matrix holders, and matrix diecases, and you try casting with an English display mould and the 49A holder, the matrix does not seat tightly to the top of the mould and the top of the body of each type is surrounded with a daisy/lion’s mane/starburst shaped squirt. If you try to continue running the caster eventually some of this gets left behind adhered to the face of the mould resulting in a much larger squirt.

To correct this, the generally accepted procedure is to place a shim behind the matrix in the holder. Up until recently I have been using two pieces of brass shim stock, one 0.010″ thick and the other 0.012″ thick, for a total of 0.022″ (0.56mm) , which seems to be enough to produce a good seal with the face of the mould. I have, however, found this to be a bit of a nuisance, as seemingly each time I change the matrix one of the shims shift out of position or sticks to the matrix. I have to store the holder with a matrix in place to prevent the shims from falling out and being lost. I have also occasionally found a matrix that needs a slightly thicker shim.

I now have an improved shim. This one is a single piece of 0.015″ shim stock, with 4 layers of adhesive-back aluminum tape for a total thickness of 0.023″ (0.58mm). I have also cut it to a shape that more accurately fits the matrix recess and also is held in, even with the holder empty, by the lugs that retain the matrix.It remains to be seen how durable this is. Oil might gradually loosen the adhesive, and the aluminum itself might wear off.

One other thing to note is that there is a limit to how thick this shim can be. Clearly if it is too thick the lugs will not be able to engage the beveled corners of the matrix, but even before that point is reached, the lugs will be held open a bit extra by the raised matrix. This will cause the Slide Retaining Pin (part code a49A4) to project from the side of the matrix holder, preventing its insertion into the carrying frame. I’m not certain what the intent of this pin is. It may be to prevent insertion of the matrix holder into the carrying frame if the matrix is not seated properly, or it might be to prevent the lugs from opening when the holder is installed (dropping the matrix onto the face of the mould) if you accidentally press the release lever.

Of course, if you are running your caster adjusted for all English moulds and matrices, the carrying frame on the bridge will be set about 0.020-0.025″ lower and this holder with work fine without any shims.

I have several of these holders so I might consider a third alternative: I could mill about 0.025″ off the rails on either edge which run in slots in the carrying frame. This would allow the holder to sit that much lower, so no shims would be necessary.

Some day I’ll spend some time measuring the vertical dimensions of all my moulds and also of the various matrix diecase types and display matrix holders, to see if I can pinpoint the origin of this height discrepancy. It would also be interesting to determine if the display holders have any vertical play to ensure that the matrix is only pressed to the face on the mould when the centering pin reaches the bottom of its stroke. Although the display matrices are not moving all over the place like composition ones, the mould blade is still opening to the set-width around the time the matrix is seating. It may be that sizing is complete before the carrying frame finishes its descent so this would be a non-issue.

More on that ‘st’ (de câlisse) matrix

My previous casting of 18-point Caslon Old Style had enough defects that I ended up re-casting it. A couple of the problems are described herein.

After calculating all the adjustments to cast the ‘st’ ligature from the non-factory-made matrix, I was still unsatisfied with the results:

(Pay no attention to the typo in the lowercase alphabet)

Even though I had its width and alignment corrected, the ‘st’ ligature still seemed to be printing soft at the top.

‘st’ on the left, capital ‘E’ on the right

Standing the type on its face showed that this sort did indeed have a slanted face: The measurement in the image are just in pixels, but knowing how deep the type was (18 points), how tall the type was (0.918″), and about how high the actual face of the type is (top of ascender to baseline, more or less the line standard 0.1591″) I can use proportional triangles to calculate that the top of the printing face is about 0.0065″ (about 0.17mm) low.

Unfortunately, correcting this is not something that can be done with caster adjustments. The mat would have to have its face and back planed to correct the angle, then built up again to get the correct depth of drive and overall thickness.

It turned out I had a much simpler way of fixing this:I had another set of matrices for quaints for 18-point Caslon O.S. which contained a factory-produced electrotype matrix in good condition (and with the correct markings). As a side note this verifies that my set-width estimate was correct.

I also ran into other flaws in the matrices for this font. At first, I thought it strange that the zero (these matrices have hanging figures) was not marked with the same width as the other figures, but a bit narrower. It also cast with a tiny overhang.

Later I noticed that the lowercase ‘o’ was casting too wide, and its matrix has a scuffed-out ‘F’ marking implying that it was a figure. Eventually I figured out that at the factory, the masters for the zero and ‘o’ had been swapped when making the matrices. The matrix with the zero had all the markings, including narrower set-width, of the ‘o’, and the matrix with the ‘o’  had all the markings for the zero, including the ‘F’ designation and a set-width to match the other figures.

Zero on the left, marked as the letter ‘o’, and the letter ‘o’ on the right, marked as the figure zero.

Just to be sure I’m not the one mixing them up, here is the sample from the specimen book:Comparing to both the matrices and my sample at the top of this post, it is clear that I don’t have them confused. The zero is an almost perfect circle of constant stroke width, while the ‘o’ is slightly off-round and has obviously varied stroke width.

A third issue which did not result in bad type but was a bit of a nuisance, was that the lowercase ‘p’ seemed to need an extra shim behind the matrix to make it seat properly on the face of the mould. Without this the top edge of the body of the type was surrounded with a sort of lion’s mane of squirt. What baffles me is that I measured the thickness of this matrix and it seems to be the same as many of the other matrices that cast just fine. I also checked for burrs on the face of the matrix and found none. The only thing left to check for (which I have not done yet) is a burr on a back corner of the matrix which could cause it to tilt instead of seating properly. My temporary shim only covered the centre of the matrix, directly over the mould cavity, so that could have seated the matrix flat.

p.s. You would need some knowledge of French-Canadian swearing to understand the title.

Exploding Mould Blade Operating Rods!

During my recent casting run of 18-point Caslon Old Style I found that there were enough problems with the font that I decided to just recast it entirely. I had solved all my alignment, set-width, and substandard matrix problems, so I thought this would be smooth sailing.

Partway through the casting, just as I shut off the pump after casting the last of a particular sort, I heard a loud clang and saw something move out of the corner of my eye. It sounded like some sheet-metal cover had fallen off near the back of the caster.

I released the clutch to idle the caster, and looked around the back of the machine. It turns out the nuts had come off the end of the mould blade operating rod, and the released force of the spring had tried to shoot all the parts across the room. They were foiled in this endeavour by an adjacent metal shelf and its soft contents (plastic bags of soda ash). The clang I heard was the parts hitting the edge of the shelf.

I spent a few minutes finding all the parts, worried that I might have to interrupt the casting job to make a replacement for an irretrievable part. One nut was on the floor directly under the end of the rod. This was probably the locknut which would have just fallen off without much flourish. The spring and inner spring seat were nearby on the floor, coated in cat hair. The rest of the parts were on the shelf, nestled among the bags of soda ash.

Nut, locknut, and outer spring seat

Remaining parts, with most of the cat hair removed, back in place on the end of the rod

I put the outer spring seat and nuts back on the rod, tightening the locknut extra tight (this has happened once before), and proceeded to finish the casting run. If it had not been for the soda ash, some of the parts might have careened into some black hole on the edge of the universe, never to be seen again.

Last time this happened I don’t think I had two wrenches to fit the two nuts and used an adjustable wrench on one of them, perhaps meaning they did not get as tight as they should have. This time I had two wrenches so I’m hoping they are tight enough to stay put.

Other than occasionally verifying these nuts are tight, I’m not sure what to do if they won’t stay on. I’m reluctant to use thread locker, as the pair of locked nuts should already be enough. One thought I had is that compressing the spring puts a bit of torque on the outer spring seat and so tries to unscrew the nuts a bit on each cycle of the caster. If the spring were wound in the other direction it would instead try to tighten the nuts.

Maybe there should be a castellated nut and cotter pin, but the rod is only ¼″ diameter so any such pin would have to be almost microscopic.

Casting #337 Caslon Old Style 18pt

This is the first time I have cast this particular font, and although it mostly went well, I had one sort that miscast. The problem first showed up on a carbon-paper proof of the type:The ‘st’ ligature seemed to be tilted off its feet so the s was barely visible. A closer look at the type itself revealed the problem:The body of the type was substantially too narrow for the width of the symbol itself, so tightening the form just pressed these beard-to-beard with no clamping of the body itself.

A look at the matrix revealed the source of the problem:

The matrix turned out to be a electro-formed duplicate made from a sample type. The matrix was originally for some letter in 14-point face #88 (Cheltenham Bold Condensed), which had been drilled out to host a new matrix. The width markings on the matrix were still those for the original sort, so I had to experiment a bit to find the correct width for the ‘st’ ligature.

I started by using a micrometer to measure the width including the beard, and did a test cast at that width. The type I measured had already had the beards crushed a bit by the lockup so this was a bit narrow and the new type still had a bit of overhang, so I widened a little more to 10 points, eliminating the overhang.

I then set a sample line, adding coppers and brasses to vary the effective width of the type, to see what width worked best. In this case the goal was to make the space between st and another t look the same as the space between two t’s.

Overinked and annotated proof

The proof revealed that the vertical alignment was a bit low, and that an extra half point of width would look good. I adjusted the vertical alignment and cast more at a width of 10½ points (wedge setting * 10 4) and took another proof.The result looked good so I cast some replacements for the type to include in the font. As a bit of validation, the width I came up with turned out to be exactly the sum of the widths of ‘s’ and ‘t’.

I’ve added an note in the box containing the matrices regarding the special width and alignment for this mat. I’m considering stamping the matrix with the correct width markings (not to mention size and face number), but first I have to decide on a good way of “erasing” the original marking. Just X’ing them out would leave little room for the new markings. Perhaps I could mill a strip off, making the matrices look like the stamped aluminum ones. I could also correct the alignment by milling a bit off the top edge of the matrix.

I was also casting some 18-point in the same face, and found that the uppercase T was casting about 0.009″ too high. This was a factory-produced matrix so I don’t know why the alignment is wrong, but the box now also has a note to adjust the alignment for this letter. This is harder to correct by modifying the matrix because it requires adding metal to the top edge.

Line Length Gauge Collection

I was cleaning up the workshop a bit last week and ran across a small project which had fallen by the wayside.

As part of all the Monotype parts I had collected, there was a random assortment of line length gauges. These are used to set the galley width on the Composition Caster to a specific measure, and to do so more accurately than by eyeballing it off the scale or even using a piece of reglet.

The set I had was very incomplete and also contained several duplicates, so the project I had set myself was to shorten some of the duplicates to fill in some of the missing lengths.

These are not the “official” Point Measure Gauge set (8CT7) that were listed with some of the Monotype tools in the English Spare Parts List. Peering at the photo in this book, it appears that the gauges they provided were in round lengths, and you were expected to stack together the correct ones to form the desired measure. The set appears to contain two 20’s, one 10, one 6, two 4’s, and several smaller ones, likely 2, 1, and ½ pica. Monotype actually produced four such sets, based on the pica standard: 0.166, 0.1667, 0.1776, or 0.1653 inches per pica, with the first being the one used for Lanston Monotype systems.

Measuring the gauges I had implied that they were made to the 0.1667 standard, so they would set the galley about 0.4% wider, or about 1 point on an 18-pica line, relative to how the rest of the caster measures things.

For consistency I chose to stick to this standard even though the 0.166-inch standard might be more appropriate with the Monotype.

The job was a simple matter of milling some metal off the end of some of the gauges, after having marked in felt tip what the desired length was.In retrospect I should have marked them in inches as well as there was one I thought I had completed but I had cut to 3.866″ rather than 3.833″. Fortunately this mistake was the right way and I just had to trim off a bit more metal.

The pieces that needed more than one pica trimmed off were rough-cut short on a bandsaw first. I had thought that a gauge that only needed one pica trimmed off could just be milled, but my mill is only powerful enough to take off 0.020″ (0.5mm) at a pass so I quickly decided that even removing a single pica needed the band-saw treatment to start.

I de-burred and polished up the gauges on my belt sander, and stamped them with their new lengths.The sanding also removed some but not all of the original stamped lengths. The new stamping was deep enough that I decided to forego grinding off the old markings completely, as it would have taken too long and removed too much metal. It turns out my newly stamped numbers were prominent enough that I think I can avoid any confusion.

Here’s the entire set, including the ones I didn’t touch, the longest being 48½ picas.

I now see I still have a duplicate 22-pica gauge, so I might shorten that one to 20. The set is still grossly incomplete, especially if half-pica lengths are desired, and I could make more from new ¼×¾″ steel stock but I’m not sure I have enough use for them to warrant the effort.

I also have one other gauge that I’m (mis-)using as a spacer when mounting an English display mould on my caster. This mould’s base is about ¼″ smaller front-to-back than the Lanston moulds, and so a spacer is required for the front mould clamp (89E1 on Lanston casters, Xa48E on English casters, and no, I don’t know which one I actually have) to hold the mould in position. I will probably find some other piece of ¼″ steel to use as this spacer (and stamp on it what it is so it doesn’t go astray), and add the line gauge (probably shortened and re-labelled) to my collection.

Casting Twisted Type

I was recently casting some 14-point Caslon Old Style on my Monotype Composition Caster and I was finding that after casting a long run of some wider letters, the type was coming out with an interesting distortion:The top of the body of the type was twisted and there was a scar on the top corner of the nick side. The damaged area lined up with the upper jaw of the type clamp, so I assume the damage was occurring as the type pusher pushed the type into the type channel and the type clamp snapped shut as the type was pushed out of its jaws. My type clamp jaws were damaged from a previous jam in the caster which probably contributed to this problem.

However, the main cause is the type coming out of the mould too hot. This was using Monotype medium alloy, which has a fairly wide temperature range over which the metal is no longer fluid, but has a pasty consistency and almost no strength. It would seem the metal was still pasty as it was ejected from the type clamp, making it easy to damage. This would also explain why it only happened for the more frequent letters in the font. After casting each letter, there would be a pause while I changed the matrix and possibly adjusted the width, giving some time for the mould to cool off a bit. As a result infrequent letters like Z would not have any problem. But for letters such as D and N which require dozens to be cast the mould would heat up as the casting proceeded, eventually ejecting too-hot type. There was no problem for narrower letters because the rate of metal injection would be lower, leading to less mould heating.

I solved the problem by lowering the pot temperature a bit and making sure I had sufficient cooling water running through the mould. I was already running the caster at the recommended speed for this size so, although slowing the caster down would have helped too, I did not do that.

This was a different problem from the usual bulging sides and/or bleeding feet that normally result from type ejecting too hot.

Making Really Red Paper

I just found these photos from a papermaking session from the summer of 2019 and thought I might post them.

We had recently emptied a mini-keg of red 112 aqueous pigment, and we were left with a dirty keg. We decided to put a batch of pulp into the keg and swish it around to use up the pigment stuck to the sides.

It turned out there was way more pigment than we expected, and we ended up with a batch of really red paper. With that much pigment, getting 100% retention is nearly impossible, so there was red everywhere. The shop looked like a murder scene (and Audrey was “caught red-handed”)!

The paper ended up being a nice saturated Christmas-y red (photos would not do it justice).

We actually used the drain water and vat water to pigment another batch of paper which ended up almost as red.

Now, after several re-uses and rinses, we can finally use the felts without getting some pink staining in the paper.

Lanston Monotype Wedge Substitutions

Based on the data from Alembic Press’s Monotype information, there are a few normal wedges that are interchangeable. So far, this is just theoretical, so I haven’t tried this. And of course, any errors in my source data will reflect as errors here as well (as will confusion on my part).

First, the following are completely equivalent:

S5 S718
S284 S77
S235B S236B

Furthermore, if you have Unit Shift available but are only using 15-row matcases, you can use Unit Shift to find the correct width from the next narrower row position for the following wedges:

Original Wedge Substitute(s)
S5 S77 S176 S284
S27 S5 S718
S29 S135
S34 S135
S77 S176
S119 S5 S77 S176 S284 S718
S150 S5 S27 S119 S718
S166 S22 S176
S176 S22
S266 S77 S176 S284 S301 S303
S283 S5 S77 S176 S266 S284 S718
S284 S176
S297 S595 S604
S298 S294
S299 S22 S176
S300 S294 S298
S302 S459
S303 S301
S314 S316
S320 S298 S300
S459 S297 S595 S604
S521 S22 S176
S606 S108 S604
S614 S200
S718 S77 S176 S284

So, for instance, if the arrangement calls for an S119 wedge, and you want to use S5 instead, when you need to cast from row 5 of the matcase (width 8 units on S119, but 9 units on S5), you instead code for row 4 with Unit Shift engaged. This will move the wedge to position 4 (width 8 units on S5 wedge) but the matcase to row position 5 selecting the correct matrix. I think this will all happen automatically if you are using the CompCat software to drive the caster and tell it you have an S5 wedge installed when you generate the ribbon file. Note that if you ribbon codes for row 15 with Unit Shift selected, the matcase may crash against the left hand side of the bridge (so don’t do that).

Finally, again using Unit Shift, you can also adjust (lengthen) the front drawbar so that the rows in the matrix case get their regular width when Unit Shift is selected for the cast, and without Unit Shift selected, you get the width from the next larger row. Note, though, that if your ribbon codes for row 1 with no Unit Shift, the matcase may crash against the right side of the bridge (again, so don’t do that).

Original Wedge Substitute(s)
S5 S27 S119 S124
S22 S176
S77 S5 S119 S124 S718
S119 S124
S200 S541 S614
S266 S283
S284 S5 S119 S124 S718
S294 S298
S297 S459
S300 S320
S301 S266 S303
S303 S266
S316 S314
S426 S283
S431 S22
S459 S302
S604 S297 S459 S595
S606 S299
S660 S606
S718 S27 S119 S124
S721 S301

This is the reverse of the first effect; To get this when using CompCat you must make up a 16-row MCA with the first row empty and the remaining rows shifted down one from their normal positions, and you must have the individual mats coded with their proper widths as if using the original wedge and a 15-row matcase.