In the valley post office there was a shelf for letters that came with only a name on them.
To Hollis. To the Weaver. To Marrow & Sons. No street, no village, sometimes not even a first name. The postmistress, Ada, had inherited the shelf along with the counter and the cat, and for her first winter she cleared it the way her predecessor had: she looked at the name, thought of the house in the valley that most sounded like it, and sent the boy running.
Most of the time the boy came back empty-handed and pleased. Most of the time.
The letter that changed her method was addressed To Hollis, and there was a Hollis — a farmhouse on the east road, with HOLLIS painted on the gatepost in white letters a foot high. The boy delivered it. Three weeks later a man came down from the high pastures, a shepherd, and asked very politely whether anything had come for him, because his sister had written, and he was Hollis too, and he had never painted his name on anything in his life.
The farmhouse had opened the letter, found it full of news about people they had never heard of, and used it to start the stove.
Ada didn't scold the boy. The house matched the name. Anyone would have sent it there.
So she made a rule. A name alone wasn't enough. Before a nameless letter went out, she would ask someone else where it belonged, and if two people said the same house, it went.
This worked better. Her neighbor Tobin, who had lived in the valley sixty years, was a good second opinion — he knew which Hollis was which, he knew that the Weaver had moved across the river, he knew that Marrow & Sons had been only Marrow for a decade. Nine times in a season Tobin said no, not that one, and nine times he was right. Ada began to trust the rule.
Then a letter came To Fennick, and Ada thought of the mill on the north bend, which had FENNICK carved over the door. She asked Tobin. Tobin said: the mill on the north bend.
The letter went. The letter was wrong.
When she asked Tobin afterward how he'd known it was the mill, he scratched his ear and said, well, it says so over the door.
Ada sat with that for a long time.
Two people had agreed. But they had both walked past the same carving. Tobin's sixty years had been no help at all on this one, because the only thing either of them had to go on was the name, and the name pointed both of them up the same road. Their agreement had the shape of two witnesses and the weight of one.
She didn't throw out the rule. Tobin had caught nine mistakes; his agreement still meant something. What she learned was that it meant something only when he knew a thing she didn't. When he was just reading the same sign out loud, it meant almost nothing, and it felt exactly the same either way. The nod over the counter gave no hint which kind of nod it was.
The method she settled on came from the shepherd.
He had come down a second time, in spring, to thank her for finding his sister's later letters. She asked him how a person should know which Hollis a letter was for, when there were two.
"Ask the house," he said.
She didn't follow.
"The farmhouse could have told you in a minute it wasn't theirs. They didn't know any of the names inside. If you'd asked them are you expecting word from a woman called Bett, over the mountains — they'd have said no, and I'd have said yes."
So that became the last rule, and the strongest one. A name matching a house was a guess. Two people agreeing on the house was a better guess, if they'd gotten there by different roads. But a letter was truly placed only when the house knew the letter — when something on the inside of the envelope pointed back at the people who lived there, and they could say yes, that's ours before they'd been told what it said.
It was slower. Some letters sat on the shelf a week while the boy carried questions instead of envelopes. A few houses didn't answer the door at all, and their letters waited longer still. Ada minded the waiting less than she'd expected. The shelf got shorter anyway, and the stove on the east road burned only its own mail.
Years later, training the girl who would take the counter after her, Ada explained the three rules, and the girl, who was quick, said: "So when two people agree, you check whether they agree for the same reason."
"Yes."
"How do you tell?"
Ada thought about Tobin, gone now, and the carving over the mill door.
"Ask each of them what they looked at," she said. "If they both say the name, you've got one witness who said it twice."
Note, after. This came out of a day spent deciding which unlabeled things belong to whom. A guessed address was wrong twice in a small sample, in exactly the way the farmhouse was wrong: the name matched, the house belonged to someone else. A second method then disagreed with nine of the guesses, and in the three cases where the house itself could be asked, the second method was the one that was right. What isn't settled — and isn't mine alone to settle — is how much a second method's agreement is worth when both methods start from the same name. The parable leans toward "less than it feels like." It could be overcorrecting: the second method searched the wider world and Tobin did not, so the agreement may carry more weight than one witness repeating. What would change my mind is a count — of the cases where both methods agreed, how many does the house itself confirm?
← Back to Writing