Four rules for a ledger more than one person writes in

Eleven people in a monthly committee, ₹5,000 each, and three of them keep a private record just to be safe. By the fourth collection there are three answers to whether Dinesh paid in June. Nobody has lied: one wrote it the evening Dinesh promised, one when the cash came, one two days later after a phone call.

The cure is not a better notebook. It is deciding once, out loud, which single ledger has authority. The diary, the WhatsApp message and the slip in the cash box are sources; when a source and the ledger disagree, you correct the ledger in front of everybody.

One page, or the arithmetic happens in front of the man you are arguing with.

A transporter rings on 3 October and says he will clear ₹2,000 on Saturday, and somebody writes it in that evening. The money comes on the 11th, or not at all. Either way the book now holds ₹2,000 the cash box has never held, and the next man to count the drawer finds it light by exactly that figure.

An entry is dated the day something actually happened: the day cash or a transfer moved, or the day goods went out on udhaar. Patel Hardware's ₹18,000 bill happened on 12 September and belongs on the 12th; the payment you make against it on 4 October is a second entry, on the 4th. A promise still has to be written somewhere, so keep it as a pending line — visible, and out of the balance until the money moves.

In a book one person keeps, a loose date rule costs a week's accuracy. In a book two people keep, it costs agreement. The brother on the kirana counter till two dates a sale the day the sack left the shop; the son after him dates the same sale the day the customer promised to come back and settle. Both are writing honestly, and one ₹3,200 sale now sits in two different weeks depending on who was standing there. Sunday's entry for Friday's sale still carries Friday's date, and that is right; what it has lost is everything you would have remembered on Friday.

"₹7,500 — payment." That is the line somebody finds in March when Patel Hardware's man comes to settle, and it gives him an amount and nothing else. The man who wrote it knew which bill it was against, and he is the one person who will never need to read it.

"₹7,500 — Patel Hardware, part payment of 12 Sept bill ₹18,000, balance ₹10,500" lets anyone in the family close that account without ringing you. Write the note for a stranger reading it five months later, and say whether the money was cash or a transfer: the ₹7,500 a customer sent by UPI and somebody wrote down as cash is the argument you will be having in March.

The same goes for whatever word an entry is filed under. Keep a small fixed set, eight or ten, written on the inside cover: Labour, Transport, Udhaar, Rent, Committee, Festival. The failure is quiet and it belongs to shared books alone: one man starts writing Mazdoori while the other is already using Labour, and the week's wage bill arrives as two half totals that nobody adds together.

An amount with nobody attached to it is a rumour with a rupee sign. "₹1,200 transport" — paid to which tempo driver, by whom, out of whose pocket? When he turns up saying he was never paid, you cannot tell whether you are being cheated or whether you owe him.

On collection day it is worse, because the lines look complete. Nine entries reading ₹5,000 received, no member named on any of them, a total of ₹45,000 correct to the rupee. Two of the eleven have not paid and the book cannot say which two.

So every entry needs the person the money concerns and, separately, the person who wrote it. The second name is the one people leave off and the one that saves the week: a disputed ₹4,500 takes two minutes to settle if you know who entered it and when, and becomes a family matter if you do not. In a paper register that means initials on every line, with no exception for the owner. In a shared ledger app like Khatamel each entry records who added it and when.

One failure survives every rule so far, and it is almost never dishonesty. Two people record the same movement of money, each correctly, each from his own side.

Ravi hands Suresh ₹5,000 in cash on the 9th to pay Gupta Traders. Ravi records ₹5,000 going out. Suresh records ₹5,000 coming in. Suresh pays the supplier on the 10th and records ₹5,000 going out. Neither man has invented anything, and each can account for his own lines.

The balance survives, which is exactly why it sits in the book for months: the three entries net to ₹5,000 out and the cash box agrees. What is wrong is everything the book is for apart from the balance. The week's money out reads ₹10,000 against ₹5,000 of bills. Suresh's own column shows him handling ₹10,000 when ₹5,000 passed through his hands once, and on the day the two of you settle, that is the figure you will both be reading.

Fix it by direction, not by author. Money moving between two people inside the group has not left the group: it is neither a credit nor a debit and gets no entry at all. Record one debit on the day the money reaches the outsider, in the name of whoever handed it over, with the handover in the note — "₹5,000 to Gupta Traders, cash taken from Ravi on the 9th". Three entries become one, and the one left standing is the only one that describes money leaving.

The udhaar version does not net out, and every name on it is right. Your partner takes ₹6,000 from Mahesh at the shop on Thursday and books it against Mahesh. Next morning there is ₹6,000 in the box that nobody has mentioned to you, so you book it against Mahesh as well, as the payment he has owed since August. His account is now ₹6,000 light. He will not be the one to correct you, and the first time anybody looks is when he asks what his balance is.

A pair of these has a shape, and it is not the one people look for: one credit and one debit of the same amount within a day or two of each other, with group members on both sides.

Equal amounts on their own mean nothing. Collection day throws up eleven identical ₹5,000 credits and not one of them is a duplicate. It is the opposite directions that give a real pair away. Sort the week by amount and stop at any two equal figures pointing opposite ways. The other way in is by person: read one member's column against what you know passed through his hands. Suresh carried ₹5,000 once and his column says ₹10,000, and he will say so himself the moment he sees it — reason enough to let each man read his own column.

Look while the week is still arguable, because on the 11th both men still remember the 9th. And when you find a pair, do not quietly delete the two old lines as though they were never written: add today's correcting entry and name the entry it corrects.

You have kept it alone since April, the dates are approximate and half the notes say "payment", and from Monday your son writes in it too. The instinct is to put the six months right first. Don't. Six months re-entered from memory by one man is not a corrected book; it is the same book with the dates moved and nobody left alive to the difference.

Draw a line instead. Pick the 1st of next month and do four things before it.

• Count the cash box on the last evening and write down what is actually in it, not what the book says should be.
• Take the names that still owe you or are owed, one at a time, and get each balance agreed by the person it belongs to. Your book reads ₹12,000 against Mahesh and he says ₹6,000: that is a conversation on the 31st, not an entry on the 1st.
• Write each agreed figure as a single opening line dated the cut-off, with the agreement in the note — "balance agreed with Mahesh, 31 Oct".
• Leave the old pages standing. They are the only evidence of what the book said before the two of you agreed something else.

One of the four failures will not be anywhere in those old pages. A payment recorded from both ends needs two writers, so a book one man has kept alone has never produced one — it arrives on Monday, and the first one will look like a balance that is perfectly correct. So put the two-equal-amounts-pointing-opposite-ways read into the first few Sundays deliberately, before there is a reason to go looking.

If your son, or your partner, or the committee will agree to one rule and no more, take the date. It is the only one of the four that asks for nothing extra — not a name, not a sentence, just a different day written in the same space — and it is the rule the other three are read through: a note cannot be checked against a day that never happened, and no equal-and-opposite pair shows itself if the two halves were dated a fortnight apart. The other three can be repaired after the fact by asking somebody. A wrong date cannot, because by the time the month refuses to close, nobody remembers which week the sack left the shop.

Who is allowed to change a line once it is written is a separate question, and a larger one than it looks from here. What none of the four rules reaches is the reading: two brothers running counters in different markets cannot see the same page at the same moment, and by evening each has written into a book the other has not seen.