August 20. The places on the equipment side were all corrected last week.
Four terminals, two reel drives, and the release-register program — eighty-six places. Correcting them took four days. The corrected places have not increased again. What increases is only the index side, and the index I have not yet touched.
At nine I ran the search. One hundred and forty-seven. On August 9 it was one hundred and fourteen, so thirty-three in eleven days. Three a day.
The guidance says to test. Its purport is that, having corrected a thing, one does not know it is corrected without trying it. That is right.
For the test I chose terminal four. It is the oldest of the four and is not much used. I pulled the line and set the clock to 1 January 2000, 00:00.
It was ten minutes' work.
The sort had inverted.
The reason can be seen. What was corrected to four digits is the place that stores, and sorting is the comparing of stored values. If 2000 goes into a corrected place while two digits remain in an uncorrected one, the two get mixed in the comparison. When the machine reads the mixture as two digits, 2000 becomes 00, and 00 is smaller than any figure.
The small stands in front. So the newest material is first.
I looked at the screen a long while. It is not that the catalogue was standing upside down. Upside down would only be the old going to the back and the new to the front, and that is simply a descending sort. Today's screen was not that. The year 2000 materials and the year 1900 materials, being the same value, sat in one mass at the very front, and behind them the twenties through the nineties stood in order.
The newest and the very oldest had entered one box. The seventy years between them queued up behind.
There are no year 1900 materials in this room. The oldest is from the twenties. So what actually sits in the front mass should be only the year 2000 items, and there are no year 2000 items either. Today is August of ninety-nine. The front mass ought to be empty.
It was not empty. There were some twenty lines. Running my eye down them, they were all things I registered after January. They were materials with an empty registration-year column. An empty column, compared as two digits, also becomes 00. Nothing, and the year 2000, and the year 1900 are one value.
At eleven I put the clock back.
Having put it back, I ran a query and the sort was as it had been. Ten minutes' work took an hour, and up to there it was all right.
Coming back from lunch I pulled the entry record. The guidance said to try both query and entry during the test, so I had made entries too. I put in three test entries and deleted them when I was done.
The traces of the three deleted entries remained. Entry records remain when deleted. The trouble was the times attached to them.
1 January 1900, 00:11. 1 January 1900, 00:13. 1 January 1900, 00:16.
Three of the year 1900 have gone into a place that cannot be corrected.
The entry record is a thing the company made impossible to correct. It is the place that keeps who put in and took out what and when, so if it could be corrected there would be no sense in keeping it. Thanks to its having been made uncorrectable, the three lines today's test left will stand as they are from now on. At the head of this room's entry record, three lines of the year 1900 have come to stand.
For six months I have tried many times only to look. To count and not increase; to read and not add. Today the company said to test, and testing is looking. Only, to look at a rollover one must make a rollover. There is no way to look at it without making one.
That is what I learned today. With some things, looking and doing are the same.
Over six months I have met that sentence in several shapes. To count one must open, and opening raises. To correct one must copy out, and copying out increases. Today: to test one must make, and making leaves a mark. The three are the same sentence, and only today's is written in the company's guidance. Whose the mark is when it is left by doing as the guidance says, I have been asking since August 2 and have not settled.
From two to four I checked the restoration. I opened the six items one at a time and looked at the times and the sorting. Terminal four was fine. One, two and three were fine. Both reel drives were fine.
When I opened the catalogue program, the sort was inverted.
The catalogue program is head office's and has never been run from terminal four. Today's test was done with the line pulled. There is no road by which something done on an unplugged machine reaches a machine that is not unplugged.
I checked three times. The line was disconnected before the test and reconnected after. Since reconnecting I have not opened the catalogue from terminal four. And the catalogue is inverted.
At half past four I telephoned the third floor and asked whether their catalogue was inverted too. He looked, then asked back whether the catalogue does not come up newest-first to begin with. I said it does not. He looked again and said that, now I mentioned it, it was odd. He said he did not know since when.
At five I telephoned head office, but the person responsible was away from the desk. They are answering telephones all day over there, I hear. Every paper is on the rollover, the government offices are circulating checklists, the banks have put notices up at the counters. With this many people afraid, what is feared ought to be distinct, and what people actually speak of is all on the stopping side. I have yet to see anyone speak of the side that does not stop but moves across.
I put down the telephone and sat and wrote out what I had done today, in order. At ten I moved the clock. At 10:11 I made an entry. At eleven I put the clock back. From two I checked the restoration. At four I saw that the catalogue was inverted.
On one machine I made a rollover for ten minutes, and it went to another machine without a line. I know perfectly well what has gone this way in this room over six months. It went to the copy, it went to the board, it went to the access record. Each time there was no road.
One thing is different. The earlier ones were all things I had not done happening. Today's is a thing I did, moving.
At six I opened the catalogue one last time. It was inverted.
I looked at the top line. There was a material number and a name. The registration-year column was empty.
The name was Pnakotic Press salvaged materials, one lot. It is the box I registered in January.