"Examiner," she said, and set the ledger sheets down first, squared to the table's edge, as Ibarra had squared her folder, as Thomas had squared everything. "The originals. Ledger sheets, eleven pages. His notebook β the accounts, all of them, in his hand. And the memorandum, with Mr. Grady's initials, which he obtained in the ordinary course of being underestimated." She did not set down the envelope. "The letter is mine. It was always mine. The rest of it is the record's, and the record is yours now. It has been in unofficial custody for thirty years and has not lost a page."
Kowalski received the notebook last, and bagged it herself, in an evidence sleeve, labeling it before she sealed it, the way Ferris had read the inventory aloud: plainly, and as if the record were listening.
"The originals will be compared against the duplicates Mrs. Vance has already provided," Kowalski said, to the room, to the signatures, to the register. "And then the scope question is arithmetic, and only arithmetic. The notebook's account numbers against the bank's surviving microfilm, account by account. If the numbers hold at scale, the restitution population is whatever the numbers say it is. Not negotiated. Counted."
She sealed the last sleeve and looked at Eleanor across the inventory table, across the empty brass drawer with its drilled lock and its thirty years of keeping, and for the second time that day her exact face relented by its one degree.
"Your husband said the numbers would keep," she said. It was not a question. She had not read the letter β Eleanor knew she had not read the letter, knew it the way you know a level is true β and yet the sentence had arrived at the examiner by the other road, the road of the duplicates and the workpapers and thirty years of a widow's testimony, and it stood in the vault between them now, a finding two auditors had reached by different entries.
"He always said it," Eleanor said quietly. "The numbers would keep."
She picked up her handbag, with the letter in it, light as a held breath.
"They kept."
The forensic accountants worked in a conference room on the fourth floor of the main branch, because the examiner's office had decided the work should be done where the records lived, so nobody could ever say the numbers had been carried off somewhere and invented. There were six of them, led by a woman named Priya Raman, who had the calm of a person accustomed to receiving bad news in spreadsheet form. They had Thomas Vance's notebook, photographed page by page under supervision and returned to its evidence envelope. They had the surviving microfilm, reel upon reel of it. And they had a whiteboard, which filled over the first two weeks with account numbers in Priya's upright hand: each one a person the bank had charged for the privilege of being robbed politely, in amounts small enough to apologize for.
The method was the one Thomas himself had invented thirty years before. He had not trusted totals; totals could be argued with. He had trusted sequences. A fee, to be a fee, had to be authorized somewhere β a schedule, a disclosure, a signature. The entries in his notebook had dates, amounts, account numbers, and beside a great many of them, in a cramped hand, the letters N/S: no schedule. No schedule, no authorization. No authorization, a taking.
Priya's team ran the sequences against the microfilm. It was slow work, and it was not, in the end, ambiguous work. A reel would go up, and the fee would be there, and the schedule would not, and a young accountant named Osei would say, quietly, "Reconciles," and write the number on the board. By the end of the third week, the board had stopped being a list and had become a landscape.
The interim finding, carried down to the examiner's liaison in a blue folder, ran to forty-one pages. The notebook reconciled. Account after account β two thousand two hundred and eighty-seven in the end, three more than Thomas himself had counted, because he had been a careful man and had still missed three β fee after fee, month after month, from the autumn of 1989 to the spring of 1997. The skim was confirmed, and it was quantified. The principal taken came to one million, nine hundred and forty thousand dollars and some cents β the cents recorded because Priya insisted that a report about stolen cents should count them. The interest owed β owed, the report said, not estimated, because the bank's own published rates for those years were a matter of its own record β came to a multiple of the principal, a figure large enough that the liaison asked whether the multiplier had been checked, and Priya said, "Three times. It is the bank's arithmetic, not ours. We only ran it honestly."