October 14. In the morning I looked at the two clocks first.
The operational machine reads 14 October 99. The test machine reads 3 January 00. The two stand side by side on one desk and their cords go into the same wall.
At nine I ran the search. Three hundred and twelve. Per the rule.
At half past nine I telephoned head office. The fourth time in four days.
I asked which machine's date the log is written by. He said the log goes by the operational machine. Then by which date, I asked, is work done on the test machine written.
By the test machine's date, he said.
Having written separate volume, I took out a new notebook. There are two unused notebooks in this room. They came together in January.
On the first page I wrote the title. Computer room log, test machine portion. Under it I wrote the date. 3 January 00.
It is the first paper in this room to carry a year 00 in my hand. Nine days ago I wrote four digits on an envelope for the first time and that was 1999, not 00. Today it is 00.
At ten the inspection sheets came down.
There are two inspection sheets.
The first is this month's. It has come down month by month for ten months, and opening six programs to confirm how they write dates and marking the sheet ends it. An hour will do.
The second is the post-rollover inspection. It says to do it within three days of 1 January 00. The test machine rolled over before 3 January, so today is the third day.
The test machine was not held in the rollover state in order to receive that second sheet. It was left so because head office said to hold it pending further testing. But the machine being in the year 00, the year 00 inspection falls on that machine. What it falls on is a machine and what does it is a person.
From eleven until twelve I did the first inspection. I marked no anomaly.
From one until three I did the second.
There are thirty-one items. Mid-October has six, so it is more than five times as many. The post-rollover inspection looks at whether the rollover went over properly, so the items are many. Date notation, sorting, leap year, calculation, entry, printing — all of it.
Writing in the separate volume, my hand kept stopping. I wrote the figures in characters. As I do in the main volume. But the main volume's date is 99 and the separate volume's is 00. The same hand writes two years inside one hour.
At half past three a requisition came from the third floor. Two boxes of second-half 1998 material.
Work like any day's. Take them from the stacks, enter them in the register, take the signature.
The register is on the operational machine. It is entered under the operational machine's date. 14 October.
Having entered it, I opened the test machine's register. The test machine has a register program too. It is a copy of the same program and the material was carried across on a disk.
There was no release for today in the test machine's register.
As it should be, I wrote, and read the line again.
There is no line, so it is as it should be. For eight months in this room what should be has several times not been. In August I saw the catalogue program turn over with no line. There was no line then either.
Writing that it is as it should be means that today that thing did not happen. There is no box in the form for writing down what did not happen. The no-anomaly mark is that box.
At ten to four I laid the separate volume and the main volume open side by side and looked at them once.
The same hand wrote them, so the writing is the same. The form is the same, so the lines are the same shape. The title differs and the date differs.
The two volumes side by side resemble what I saw yesterday. Yesterday the index set two lines in one place across sixty-two years. Today I set two volumes on one desk. What set them is not the index but me.
Only what I set out is two papers, and the years the two speak of are two. One is 99 and one is 00. Not sixty-two years but one year apart. The one year in between has come inside this room.
Whether I brought that year in or not will not be told apart. Head office ordered the test and head office ordered the rollover state held. Head office ordered the separate volume too. There is someone who ordered it and the hand is mine. In August I stood in that place once and wrote something down then. Is it different when there is an order.
From four until five I did cards. I finished drawer seven and opened drawer eight.
The cards are one set. There is no separate set of cards for the test machine. Cards are paper and have no clock. What has no clock does not grow two years.
What lives two years in this room are the things with clocks. The catalogue and the register and the index and the access record. All four are records and all four have a date box. Where there is a date box, which year goes into it the machine fetches from its own clock. Two machines, two years.
At five I counted what I had done today.
Nine lines in the main volume and seventeen in the separate volume. Twenty-six.
Yesterday it was thirteen. The day before, twelve.
The objects have not increased. The programs are six and the drawers are twelve and the stacks are as they were. What increased is a date.
One date increased and the work doubled.
Nine days ago I wrote in this log that the work is shrinking. I reckoned that correcting outruns gaining, and that what is gained comes already corrected, so my share is only the original seven thousand three hundred and forty. That reckoning still holds. The cards are one set and they are ahead of schedule.
The reckoning holds and the work is double. It came from where the reckoning was not. What I counted was cards to correct, and what increased today is lines to write.
At half past five I thought again about which clock I ought to work by.
To settle is to throw one side away. For eight months I have always settled. I settled on believing the log over the copy and the paper register over the screen. Settling is what makes work possible even with no grounds, so I settled.
Today's reply says not to settle. It says to use both. Not settling has entered the form for the first time.
Not settling, two years' worth comes in. Today it is twenty-six.
At twenty to six I pulled the duplicate-entry list. It is the fifth list, made on 6 October.
There were sixty-one lines.
The count of date fields I converted from two digits to four since August is in the log. I looked out that figure.
Sixty-one.