Cooking paste badly, on purpose

Every bindery has a paste recipe and every bindery is certain theirs is correct. Ours is one part wheat starch to five parts water, cooked twenty minutes, strained through a cloth. We have never had a reason to question it, which is exactly the problem.

So we cooked four batches: one to five, one to seven, one to nine, and a deliberately undercooked one to five. We pasted matching strips of Japanese tissue onto offcuts of the same handmade paper, dried them under weight, and left them on the shelf for six weeks.

The undercooked batch looked fine on day one and was powder by week four.

The thin one to nine mix behaved better than expected. It went on almost invisibly, and the tissue stayed down. Where it failed was time: two of the strips lifted at the corners, which is what you would predict, but they lifted cleanly and without taking any fibre with them. For a repair you want to be able to reverse, that is arguably the right kind of failure.

What we changed: nothing about the standard recipe, but we now cook a thinner batch alongside it whenever we are lining something we expect a future binder to have to undo.