Now serializing

The Number Beneath the Ledger

Autumn 1997, the records room of a merchant bank. In the hand that copies old ledgers, the numbers grow as they are set down, and the defaults arrive first, dated to a day not yet come.

Read from the start Latest · ep. 104

PNAKOTIC PRESS

Recent chapters

104
More to Correct
July 10 (computer room log, page seventy-two). The tomorrow written in yesterday's reply is today. The log's page seventy-two is empty down to the page number; the copy has it. Today the narrator looks at the reply's poster name, unread yesterday, and finds the four digits keyed into the name field when the account was made. The account page holds two posts (correction, reply). The access record shows creation at 16:20, posting at 16:51, logout at 16:53 and nothing after; there is no access corresponding to the reply's 17:40. It is the digital counterpart of the place where the predecessor wrote that the hand was his own and he had no memory of writing it. Only a hand can be imitated and a name is four digits: too simple to call imitation, and of all names one made yesterday, which is too much to call coincidence. Keying in the old password before changing it, it goes through unaltered. The reply deletes — and that it deletes, where the line in the copy would not, is the stranger thing. After deleting, a search finds the three lines already reproduced in two places: one set among other passages with no mark where the insertion begins, one moved whole with a question mark at the end, asking what the passage means. Only the narrator can answer, and answering adds a copy. What was deleted is one original; the two reproductions stand. Yesterday's correction has itself been copied four times, and in one of the four a unit in the pit's recorded depth is a character off. The original put up to set something right has spread wrong in turn: setting right makes the next thing to set right. Working out the other side fails, because how long the first sentence took to go from one copy to sixteen was never counted, and counting would have meant opening. At three the narrator posts a second correction and attaches the address of the first, then understands that attaching an address means more people opening it. In the afternoon someone down from the third floor glances at the screen and asks whether ours is on there; answering no, the narrator works out which copying-out that answer made today, since speech goes into the count once it is copied out. The four o'clock search returns twenty-four against yesterday's twenty-one, and two of the three additions are reproductions of the correction. One added yesterday became three the next day — the feedback is established. End hook: after finishing the log entry, the copy shows page seventy-two filled at twenty-four, and beneath it pages seventy-three, seventy-four and seventy-five sitting all at once, dated the day after tomorrow and the two days following, their figures empty. What ran one page ahead until yesterday ran three pages ahead today.
- 1999-07-10
- 3,512 characters
- ~10 min
103
The Line I Corrected
July 9 (computer room log, page seventy-one). Opening the copy first, the empty figure on page seventy has been filled: twenty. This time the narrator sets the order, holding out two hours without counting, to measure how long the not-knowing can be held. The 11:14 a.m. search returns eighteen. For the first time in six months the copy and the count come apart, and the narrator feels something like relief. The 2:00 p.m. recheck returns twenty; the two additions were posted at 11:50 and 12:40. It had not come apart. It was not yet. The falsification procedure collapses here: separating wrong from not-yet requires fixing a term, no grounds exist for fixing one, and whatever figure is written is passed in the end. A table that cannot be wrong is not a table that is right. Meanwhile the line of unknown origin in the third copy, a sentence saying the surveyor left the pit after his last measurement, has propagated to sixteen of twenty. It is absent from the original, and six of the sixteen carry a remark beneath it; the word relief appears three times. With that sentence the story closes; without it, it does not. The closing kind is easier to copy out. Correcting it means writing the original out, and the typed kind grows faster. Leave it and the sentence spreads; correct it and one more copy of the original spreads. Working out which adds less requires the rate, the rate requires counting, and counting puts it up. At four the narrator makes an account on an external board, keys the four-digit archive code into the name field, and recognizes those digits as the ones sitting in the recipient field of the last log entry the predecessor wrote. They are not deleted. Half of the correction is quotation. After posting, the search returns twenty-one; the one added went out from the hand of the narrator, under a name of a name the narrator made. The one who counted for six months has put a line into the side that gets counted. The surveyor only measured, and measuring is not laying a hand on the place; today a hand was laid on it (the propagator’s recognition crosses into act). End hook: one reply under the correction. Page number seventy-two, a date that is tomorrow rather than today, two figures, one twenty-one and the other empty.
- 1999-07-09
- 3,487 characters
- ~10 min
102
Page Seventy
July 3 (computer room log, page sixty-nine). With nothing settled to write, only the page number and the date go down and what comes under them is left empty. Leaving it empty was supposed to mean not writing, but the page number is already written — and the three empty pages in the predecessor's notebook, read as a break in the work while scanning, turn out to have numbers too. Four days of holding out end on July 7. Those four days were spent thinking about the line, which is to say reading it without opening it. The empty second figure has been filled: fourteen. The search that follows turns up fourteen locations, where five days earlier there were eleven — all three additions of the typed kind, two of them copies of someone else's copy. The order has reversed: in June the counting came first and the figure went into the table; in July the figure was there and the counting went to confirm it. Two readings split — someone counted first and wrote it down (leaving open why it carries the narrator's own log page number), or the figure was written and then there were fourteen (no way to explain it). The only way to separate them is not to confirm, and the confirmation is already made. The copies grew over four unopened days, so the narrator is not the only cause; but four days they did not open are four days others did, and there are no four days nobody opens. Read a third time, the survey log shows something else: the man knew what he was doing and kept writing it down, and a man who measures, if he stops measuring, stops being one. The turn: the log has never been posted anywhere and sits in a locked drawer, yet the copy carries its form and follows its page numbers. If the copy a hand passed through gains more, the hand laid on this material most often is the narrator's — so it is not that they copied it out, but that they are being copied out. At the end, page seventy is written with the next round's figure left empty; opening the copy once more, page seventy is already there below. One figure is fourteen. The other is empty.
- 1999-07-03
- 3,706 characters
- ~11 min
101
The Second Column
July 2 (computer room log, page sixty-eight). The day the empty column ruled a month ago gets its second entry. The thing that arrives with the table open: to count them I have to open them, and opening puts the visitor figure up by one. Through June the table was to stay shut, yet the log counts twenty-three openings, and no reason was ever written down — that file does not even sit beside other files, its registration number is nowhere near them. Going to find the seven, there are eleven, the four new ones all private. No copy has fewer lines; what differs this month is the size of the gain, and the copies whose views rose most gained most. The hand stops at our registered copy. It exists on the internal network alone and cannot be opened from outside, and for the month of June the only person who opened it was the narrator (forty-eight to seventy-one internal views, twenty-three of them theirs) — and that file gained a line. The twelfth line sits in what had been margin below the lighthouse log's tide table: one tide, its date falling between the dates above and below, a missing day filled in. Nothing strange in that reading, except the file was made by photographing the paper, and on the paper that place was margin on June 1 and is margin today. The two readings that were supposed to separate when measured at two points did not separate. To claim the reading grew it, one copy would have to go a month unopened — but confirming that an unopened copy has not grown means opening it. There is no control to be had. The survey log's table, the same place measured over and over with the figure growing one way, had been read as error while scanning; wavering has no direction, and those figures had one. Measure it and it deepens, set down here, becomes: count it and it increases. Going round the eleven again in the afternoon, each visit enters next month's third column. The largest gain belongs to the homepage where all hundred and eighteen pages were keyed in character by character — the copy a hand passed through gains more than the copies posted as photographs. At the end, three of the four new lines are tides, depths, something lit at night; the fourth has a date and a page number and sets down figures without judgement, and the narrator knows one person who writes that way. Their log stands at page sixty-eight; the line gives page sixty-nine, and of the two figures that follow, one is eleven and the other's place is empty. The registered copy is on the internal network, the log is in a locked drawer, and one person has seen both. The sheets in the drawer make four, and today all four look like sheets that record one thing.
- 1999-07-02
- 3,772 characters
- ~11 min

All 104 chapters →

Completed

All series →

Recent chapters

More to Correct

The Line I Corrected

Page Seventy

The Second Column

Completed