Six copies of a poetry pamphlet. 128 pages in sixteen sections of eight, trimmed to 216 by 138, cloth over 2.0 mm board, sewn on three tapes. Five are on the shelf behind me and they are good books. The sixth went in the bin, and the reason is stupid enough that it took me thirty eight years to get round to it.
| Item | Spec | Note |
|---|---|---|
| Text paper | cream wove, 120 gsm | grain head to tail, no exceptions |
| Board | greyboard, 2.0 mm | arrow the grain before it goes in the rack |
| Cloth | paper backed book cloth | PVA only; paste cockles the backing |
| Thread | linen, 18/3, beeswaxed | thicker than 35/3, not thinner1 |
| Tapes | cotton, 13 mm, starched | three, cut to 120 mm, ends under the pastedowns |
| Linings | mull, then kraft at 90 gsm | kraft grain head to tail |
| Hinge tissue | kozo, 9 gsm | only for a fragile first leaf |
| Adhesives | neutral pH PVA; wheat starch paste | not interchangeable2 |
Grain before anything else touches the paper. Flex a sheet between both hands, turn it ninety degrees, flex again: the direction with less resistance is the grain. In Dutch it is the draadrichting, the thread direction, which is the better name, because fibres laid down on a moving wire remember it forever. Text paper, endpapers, board, cloth and linings all run head to tail in the finished book. Two sheets folded once and nested gives an eight page section, one pass of the bone folder from the middle outwards.
Sections into the standing press between boards overnight, because a folded section is thicker at the fold than anywhere else. Mark the stations once, on a card cut to the spine length. Kettles 15 mm in from head and tail, three tapes across what is left: on a 216 mm spine that puts tape centres at 61.5, 108 and 154.5, and a 13 mm tape wants holes 6.5 mm either side of centre, so 55, 68, 101.5, 114.5, 148, 161. Eight stations, punched from inside the fold with a bradawl.
18/3 linen, run twice over a block of beeswax, three metres at a time and rejoined at a kettle because a long thread fights you. Every time the needle comes out at a tape station it crosses the outside of the tape and goes back in the other hole. Over the tape, never through it; the tape is what carries the boards.
Sixteen sections sewn with 18/3 adds something like 6 to 8 mm at the spine over the fore edge. That is the swell, and it is not a fault. It is the material you round with. Want less of it, sew 25/3 and accept a weaker book.
Block into the finishing press, spine standing 5 mm proud, knocked square against the cheeks and checked at head and tail. PVA thinned with a little water, brushed across the spine rather than along it so it goes into the folds. One coat, wiped back with a fingertip so no bead sits on the surface.
This is the step you cannot take back. An old animal glue spine softens under a damp poultice and the sections come apart intact. A set PVA film does not; it comes away with paper attached. Anything wrong with the collation, the sewing or the squareness has to be found before the brush goes in.
Round first. Spine towards you, left hand pulling it over while the right taps the top third with the flat of a backing hammer. Turn, repeat, small passes. About a third of a circle, and the useful check is the fore edge, which should have gone concave by as much as the spine has gone convex.
Then back, in the press between backing boards set so the spine standing above them is exactly the thickness of the board that will sit in the shoulder: 2.0 mm each side. Not 2.5 to be safe. Glancing blows at a shallow angle, from the middle out to each shoulder. You are not hitting the book. You are persuading five sections to lie down.
Mull, 20 mm shorter than the spine and 25 mm wider each side, PVA, rubbed down through the round so it takes the shape instead of bridging it. Then kraft at 90 gsm cut to the exact spine width and height, grain head to tail. One layer on a 128 page book, two if it were 300.
A folio of the same cream wove, folded with the grain, tipped on with a 3 mm bead of wheat starch paste along the fold. Three millimetres, not five. A wide tip line stiffens the first leaf and the book stops opening flat at the front, which accounts for more disappointing amateur work than anything else I see.
2.0 mm greyboard, grain head to tail. Height is the text block plus a square at head and tail: 216 plus 3 plus 3 is 222 mm. Width is the text block less the joint gap plus the fore edge square: 138 minus 8 plus 3 is 133 mm. Check every board against an engineer's square, because a chopper half a degree out gives you a case that will not close and you will spend an hour blaming the cloth. @rustbelt/scraping-a-1943-lathe-bed-flat is the best explanation I know of why you check against the square and never against the last thing you cut.
Cloth: two boards at 133, 36 across the round of the spine, two joints at 8, and 15 mm of turn-in each side. 348 by 252 mm. Corners cut off 3 mm out from the board corner on the diagonal, that being one and a half times the board thickness.
Case open and flat, block laid in on its back board, positioned by the joint and not by eye. Paste the endpaper, close the board onto it, rub down through siliconised paper from the joint outwards. Turn over, repeat. Nipping press with brass edged boards seated down into the joints, thirty seconds of real pressure, then waxed paper between each pastedown and its board, under 20 kg overnight.
What I check in the minute before step four, off a card that has been on the wall since about 1994:
That last box is not decoration. It is the one I did not tick.
On copy three I got clever with the board. Eleven boards were already off the good sheet. The twelfth wanted 222 by 133 with the grain along the 222, and the only piece left in the rack was a 205 by 380 offcut with the draadrichting going the short way. A 222 mm board will not fit along that grain. It fits beautifully across it. Four minutes down the stairs for a new sheet, or one rotation of the board and I go home. It was a Friday, ten past five.
Cased in, nipped, waxed paper, twenty kilos, lights off.
Saturday morning the back board had cupped away from the block. Steel rule from joint to fore edge, feeler gauges stacked under the end of it: 3.8 mm at the fore edge corner, tapering to nothing at the joint. The front board, cut the right way from the same batch with the same cloth and the same paste, measured under 0.2 mm, which is what I call flat on a cloth case.
No mystery in the mechanism. Glue and paste put water into paper, paper swells as it takes water up, and it swells considerably more across its grain than along it. Drying, it contracts back and takes whatever it is stuck to with it. In a properly made case the cloth outside and the pastedown inside pull on the same axis, and because the board's grain agrees with both, the board is asked to bend about its stiff axis and refuses. Turn it ninety degrees and that contraction acts where the board has no stiffness at all. It goes like a crisp.
Then the second mistake, which was trying to save it. Damped the cloth, 40 kg with a slightly convex board on top, three days. It came out at 1.4 mm and I was pleased with myself. A fortnight later it was back at 3.1 mm. The board is not warped. It is doing what a board with the grain that way does under a pastedown, and it will go on doing it long after I have stopped being able to explain myself.
New board, new case, two hours of a Saturday, and the sheet of greyboard I did not want to open costs about two euro fifty. Every sheet coming into the shop now gets an arrow in carpenter's pencil along the grain the day it arrives, offcuts included, before it goes in the rack. Ten seconds, and the decision gets made by somebody who is not tired.
The numbers on linen thread are a count, not a diameter, and they run backwards: the first figure is roughly how many 300 yard skeins go to the pound, so a higher number is a finer thread. 18/3 is three plies of an 18 count and is thicker than 35/3. Measured, 18/3 comes out near 0.6 mm and 35/3 near 0.4 mm, but one maker's 25/3 is not another maker's 25/3, because twist and dressing vary. Buy one brand and learn its hand. ↩
The difference is water. Paste is mostly water, which is why it gives you working time, why it cockles paper, and why conservators use it: a starch paste joint releases with a damp swab, so a repair made this morning can be undone by somebody in 2090. PVA carries far less water, sets fast and does not release. So PVA for cloth to board, paste for the pastedown. Wheat starch and not flour, either: the gluten in a flour paste cross links as it ages and becomes very hard to remove. ↩