October 20. In the morning I went down to the stacks first.
There is one gap in the first-half 1998 section and a box beside it. The same as thirteen days ago. It is the box the register has entered as going out today.
At nine I ran the search. Three hundred and thirty.
Having written no requisition received, I waited out the morning.
While waiting I finished the count. In six days I did it twice more.
Sixty-eight and sixty-eight.
Agreeing once is coincidence. Agreeing twice is coincidence twice. At the third time the first thing to doubt is the way of counting. So I wrote down first that the one who counts the two figures separately is myself alone. The counter being one, I may be counting the two as the same thing.
There was one way to show that I am not counting them as the same. Matching them by date.
From half past ten until one I matched sixty-eight pairs by date.
One to one.
Three fields converted on the eighth, and three duplicates arose after it. One converted on the ninth, and one arose. In fields not converted, not one arose.
From one until two I looked at whether this agrees with the explanation head office gave.
Head office's explanation is this. Where the converted side and the unconverted side say different years about one material, the line becomes two. Convert them all and it goes away. When the conversion finishes at the end of November, guidance for clearing them comes down.
By the explanation, duplicate entries do not follow the number of converted fields. They follow the number of places where converted and unconverted meet. Places where they meet increase by multiplication. Ten converted and twenty not gives two hundred meeting places. Twenty converted leaves ten, so two hundred again, and twenty-five leaves a hundred and twenty-five.
It ought to increase by multiplication and then fall. It rises quickly at first, begins to fall past the halfway mark, and comes to nought when all are converted.
Sixty-eight is not that shape. It rose one for one and has not once fallen.
Mechanism and measurement differ, I wrote.
The explanation went as far as nine again. Why duplicate entries arise, the explanation has right. Opening them one by one, every one has the converted side and the unconverted side saying different years. Where the explanation does not go is the figure. Why it is exactly that many is not covered.
From two until three I sat over how this figure was to be written down.
One converted field, one day gained.
For eight months in this room I have met this shape many times. In February, lay a sheet of paper on the scanner and one file came into being. In March, copy one file and the copies increased. In June, correct one wrong line and post it, and the corrected text spread wrongly again and there was more to correct. Each time I chose to put a hand in.
In August it was not so. Head office sent the rollover work down to this room and I put a hand in as told, without choosing. There is something I wrote in the log then. Is it different when there is an order.
Two months later the answer came. It is not different.
That there is someone who ordered it takes nothing away.
I thought something would be taken away. Six months in this room taught me that putting a hand in makes things grow, and having learned it I spared my hand. I tried what would not work three times and stopped, and I left a whole day uncounted. Sparing was the work.
When the order came in August I thought a place had arisen where I need not spare. I held that choosing was what the cost attached to. The cost attaches not to choosing but to putting a hand in.
At ten past three I opened the separate volume and counted the test machine's figures too.
The test machine has no duplicate entries. Its programs are all converted and there is no unconverted side. With no unconverted side there is nothing to say a different year against. Just as head office explained.
This is a place where the explanation is right. Finding and writing down places where the explanation is right has happened a few times in eight months, and each time I was eased a little. Today too.
Having been eased, I looked at one more thing. The test machine has no duplicate entries and instead has one more date. The years that machine lives have increased by one, and the work has doubled.
On one side there are lines gained and on the other a day gained. There is no machine in this room with neither.
At half past three I went down to the stacks. The box was there as it was.
At four I pulled the register. The 20 October line was there as it was. There is no release today.
At twenty past four the third-floor man came down. He came without a requisition slip.
Looking at the catalogue upstairs just now, he said, that box is down as going out today, so he came down. It is handed out when there is a requisition, I said. He said he would write one, then. He wrote it, signed, took the box and went up.
Before he went up I asked one thing. Was the box needed today.
It was not needed, he said. It was down in the catalogue as going out today, so he came to have it put away.
He came because it was written.
From ten to five until five I sat with that.
Thirteen days ago today's date was entered in the register. Today that box went out. The register was right.
The reason it was right is two things. It was right because the register wrote it first, or it was right because he was a person who would take that box today. By the second, the register only knew in advance, and that is what the copy has been doing for eight months.
By the first it is otherwise. He came because he saw the catalogue. That line has been in the catalogue since thirteen days ago. Had the line not been there he would not have come down, and it was not needed either, so the box would have stayed on the shelf.
For eight months what this room's records moved were figures and paper. View counts increased and cards increased and lines increased. Fourteen days ago what increased was an act, and writing it down I wrote that for the first time in this room work had increased.
Today it is a person.
One person came down the stairs and took away one box. The reason he came down was written on paper. Not need.
At ten past five I pulled the register for the last time. It is a habit, checking the day's.
Lines coming up under that material number were four.
First line, 6 October 1999. Second, 6 October 99. Third, 20 October 1999. What went out today is the third.
The fourth line was 31 October 1999.
That is the date of the two unexplained items I saw in the catalogue thirteen days ago.
I went down to the stacks. There were two gaps in the first-half 1998 section.
There is no box to hand out on the thirty-first.