April 7. I started making a comparison table.
It is work I put off through March because other work piled up. Enquiries about the old logs kept coming in after the newsletter, and each time I opened the file and answered. Answering, it crossed my mind now and then to check whether anything differed from the original, and there was no time.
In the first week of April requests fell off. A month after the newsletter the enquiries thinned out too. That is when I started.
The method was simple. Put the file on the screen a page at a time, lay the original beside it, count lines. If the counts match, move on; if they differ, note which line. Six hundred-odd pages over five bundles takes days that way. I decided on an hour a day.
Nothing came up through the fifth day. A few pages had different line counts and all of them were my mistakes — counting a fold as a line, or a fingertip photographed in the top margin as a line. Comparison is dull, and dull work puts mistakes first.
On the eighth day page 112 of the lighthouse log came up. A place I already knew. After that they came often.
On April 20 I finished all six hundred and fifteen pages.
There were eleven lines present only in the files.
Three in the lighthouse log, two in the survey log, three in the telegraph log, two in the stacks log, and one in the archive record. The line that ran in the newsletter was one of the lighthouse log's three.
Having confirmed all eleven, I copied them out onto paper. I copied them because on screen they were scattered and did not come into view at once. I only meant to have them on one sheet.
It took about ten minutes. The lines are short, so they fitted on one sheet.
When I had finished writing I looked at the sheet for a long time.
The eleven lines ran on.
Each line had been on a different page of a different bundle. Page 112 of the lighthouse log and page 80 of the survey log have nothing to do with each other. They were written sixty years ago by different people in different places. And set down in the order they were found, they were one sentence.
I did not read the sentence aloud. Reading it felt like conceding that it is a sentence. Instead I changed the order. I shuffled the bundles and shuffled the pages and wrote it out again.
Shuffled, it was not a sentence. It ran on only in the order it was found.
I checked once more. My hand shook a little while I checked, and the reason was less fear than not knowing how this was supposed to be written up.
Set out in order, where the eleven lines sat showed no pattern. Lighthouse 112, then survey 80, then telegraph 29. Forward, then back. Only the order of the bundles ran always one way. Lighthouse to survey, survey to telegraph, telegraph to stacks, stacks to the archive record. From the people of sixty years ago toward last year's person.
The run-on sentence, placed after the sentence that breaks off on the archive record's last page, continued it exactly.
From the one who reads to the one who copies, from the one who copies to. That is what is on paper. Put the eleven lines after it and the sentence ends.
I will not write what follows here. If I had to give a reason it is that it is not the sort of thing for a work log; exactly put, I did not want to write it.
I tried to confirm when they appeared.
A file carries the time it was made. Look at when it was first saved and you know from when the line was there. I opened it thinking that, and there was a problem.
There was not one copy.
One I registered, one on the repository server, one converted copy sent to audit, and after the newsletter some two hundred downloaded by departments. The times made are all different. The content is identical.
By time, mine is first. Then the line was there from when I scanned. But when I scanned I saw only the one stroke on the lighthouse log's last page and did not see the other ten.
Not having seen and not having been there I cannot tell apart. Two hundred pages a day and the eye goes past. Until last month that is how I settled it.
Only, the newsletter line is different. Planning's writer said they read all hundred and eighteen pages. If someone who read it all picked a good line, then the line was already there when they picked.
So it was there at the start of March. That line, at least.
Whether it was there in February there is no way to confirm. No one read that file end to end in February.
I thought briefly about whether I could ask the computer-room people. Whether writing that was not in a file can come to be in it. Ask and I know the answer. It cannot. A file is fixed when it is made and after that it is only opened. Saving stamps a new time, so if someone edits it, it shows.
It should have shown and it did not. Then no one edited it. And if no one edited it and the writing grew, I cannot make the next sentence myself.
On April 21 I opened the box again.
I opened every page the eleven lines belong to. They were not on the paper. All eleven places were margin or another sentence.
The original is as it was. I confirmed that and closed the box. Confirming took an hour and I looked at all eleven places twice.
Closing it, a strange relief came. If the paper is as it was, then what grew is the file side, and files are the company's, and what is the company's is managed by regulation. What regulation manages is not my responsibility.
How thin that relief was I learned that evening.
Tidying the comparison table before going home, I looked again at the sheet with the eleven lines. The sheet I wrote out by hand this afternoon.
That sheet was now the only object in this company where that sentence runs on as one line.
In the files it is scattered. On the paper it is not there. The only thing that holds it joined is this one sheet I made.
I put it in the drawer and locked it. Locking it, I decided not to think about what locking changes.
The connection log I printed last month was still in the drawer. The two sheets went in side by side. One is a record that a machine with its power off attaches every night, and one is a record that a sentence that was not there runs on.
Set side by side they look related. Looking related and being related are different. I was working to keep that difference, and I knew that working at it already meant I had leaned one way.