October 13. I took out the box from under the desk.
I came to this room in January and opened that box, and I have opened it some eight times since. Each time I opened it as material. What is inside is four bundles and the predecessor's note. The four are old logs with titles on them and no registration numbers. It is a box not on any list, so it cannot be registered, and not being registrable it is not material. Eight times I have opened what is not material.
Today I took out the third bundle. It is a telegraph log.
At nine I ran the search first. Three hundred and nine.
Leaving the material number box blank is a first in eight months. This room's form is made so that what has no number is not opened.
The telegraph log is of the year thirty-seven. The year stands on the cover in two digits. 37.
There is a person inside it. Working alone at night, he was listening to a frequency nobody calls. Answers come on the frequency that is not called, he wrote down. He also wrote down that he had been told not to answer. AUDITUR QUOD NON SONAT.
Through eight readings I read that passage as sentences. Today I opened it to look at the time box.
In the last five or six lines there are receipt times.
They are written in four-digit numbers. 0251. 0259. 0300. 0304. 0311.
Five of them.
All five are the same.
When I saw the five entry times yesterday I recognised four from the access record, and one was left over. Today five came out of one place. It is a thing of sixty years ago.
From half past ten until twelve I worked out whether this is a thing that can happen by counting.
There are one thousand four hundred and forty times in a day. From ten minutes to three until a quarter past three there are twenty-six. The likelihood of five falling inside them can be counted. Counted, it is small. That something is small does not mean it does not happen, as I wrote in June.
Only the five are the same in order as well. To cover order too by counting, one must count every way of choosing five out of twenty-six and setting them in a row. That number runs to seven figures.
Insufficient, I wrote. A small likelihood and a thing that did not happen are different.
After lunch I opened the index. There was one thing to verify.
How the index stores a time.
The time stands apart from the date. It is one four-digit number. Whether the year box is two places or four, the time box does not change.
So 0300 is one value. Three o'clock of 1900 and three o'clock of 2000 and three o'clock of thirty-seven sit down in the same value.
I put 0300 into the index query.
Two items came out. One line of yesterday's register and one line of the material with no registration number. That something without a registration number comes up in a query I also saw today for the first time. It is because I opened it in the morning. Open it and it comes up.
The two lines stood side by side in one place. Sixty-two years lie between them and on the screen those sixty-two years are not there.
I looked at the two lines a while. I had to know what their standing side by side does.
An index is a thing that gathers like values. Where values are the same it gathers them. That is the whole of what an index does. What set the two lines side by side is the index and not the two lines.
But the value an index gathers by is settled not by the machine but by the form. It was people who settled that the time stands apart from the date. Standing apart, a time loses its year. A time that has lost its year becomes the same value across sixty-two years.
For eight months in this room I have several times seen two records stand side by side. The copy and the log stood side by side and the paper register and the screen stood side by side. Each time I asked which came first. Today there is nothing to ask. A time has no before and after. 0300 was 0300 sixty-two years ago and will be 0300 tomorrow.
The form has made one box with no before and after in it. It was not made to save a place and it means nothing at all. A time repeats day by day, so there was simply no reason to keep it fastened to the date.
From two until four I read that passage of the telegraph log again.
What the operator took down was not writing. It is a row of figures. He wrote that he did not know what it meant. Five groups of digits stand in one line and below them are three question marks. Places cut off during receipt, it says.
I counted the places in the groups.
Four. Eight. Five. One. Four.
It is the shape of one line of the release register.
Only the place counts are the same. The figures cannot be matched. Three places in the eight-place group are illegible, so the material number cannot be matched, and there is no material to match it against. The eight-place registration numbering came into being last year.
At four the third-floor man came down and took a signature. On his way out he said I stay late these days. I leave at six, I said. He said the light was on yesterday.
I left at six yesterday, I said. He said he must have seen it wrong, then, and went up.
After he went up I did not write it in the log. There is no box for it. There is no box in the form for what someone else saw, and had there been one, I left at six yesterday.
From half past four until five I sat over how today's matter was to be written.
There is one thing to write. The shape of the figures taken down sixty-two years ago is the shape of one line of this room's register. Then what the operator took down may be this room's record.
To write that is to write the reverse as well. This room's record may be following a shape that was taken down sixty-two years ago. By the first, this room was there before. By the second, that night was there before.
There is nothing to tell them apart by. A time is one value and has no before and after, three places of the figures are illegible, and the places to ask have gone one by one over eight months.
Having written that it is entered in this log only, I read the line again.
Among the things eight months in this room have taught me is that writing makes things grow. The form had no box, so I wrote it in the log only. The log is this room's record. If what was taken down sixty-two years ago is the shape of this room's record, then this line I am writing now is inside that shape too.
At half past five I closed the box and put it under the desk.
Before closing it I looked once at the predecessor's note. The last page breaks off in the middle of a sentence and my hand carries on after it. After that there are two blank pages.
At thirty-two minutes past five I looked at the two machines' clocks side by side.
The operational machine reads 13 October 99, 17:32.
The test machine I left as it was, head office having said to hold the rollover state pending further testing.
It read 2 January 00, 17:32.