In March, Della sat at my desk with the guide open and asked me not to point. I had already lifted one hand toward a drawer. I put it in my lap and moved my chair back far enough that she could open the drawer herself. We had chosen an ordinary payment-record check, the sort of task I could complete while answering a question about something else. That was why I thought it would be a fair first trial.
She found the folder, located the entry, and compared it with the supporting page. Then she stopped. There was a note beside the entry directing her to a separate sheet, but the guide described no separate sheet. She turned back one page and then forward two. I knew where the sheet was. Knowing made the silence feel wasteful, and I began to lean toward the desk.
"What would I do now if you weren't here?" she asked.
"Check the attachment behind the monthly divider."
"Which instruction tells me that?" She was not irritated. That made it harder to defend myself. I took the guide when she offered it and found the sentence I had written: compare supporting material. It had seemed plain when I wrote it because I could picture every piece of material involved. Nothing on the page told another person what I was picturing.
I explained that this entry carried its supporting detail on an attachment instead of the main sheet. Della asked how she would recognize another such entry. I started to say she would get used to them, then stopped. That was how I had learned too much of the work: by repeating an uncertainty until familiarity hid it. I asked her to leave the folder as it was while I wrote down the missing step.
The first revision was too long. I described three possibilities she had not encountered and buried the useful direction among them. Della read it and tapped the middle of the paragraph. "Is this what I do first?" I crossed out the opening and put the action first. We added where the attachment belonged, what the notation meant, and a place to stop and ask if the papers did not match.
"Shall I finish this one?" I asked.
She moved the guide closer. "Let me try it again." I sat back. She returned to the point where she had stopped, found the attachment, and completed the comparison. It took longer than it would have taken me. It also left us with something we had not possessed that morning: an instruction that had survived being used by someone who had not written it.
Through April and May, we repeated the trial with other ordinary tasks. I kept a page for corrections instead of silently changing the guide and presenting it as though it had always been clear. Some questions concerned missing words; others showed that I had combined two decisions because I usually made them together. I began to measure progress by what could proceed without my hands, not by the neatness of the binder.
At the garage in June, Otis was fitting a part back into a tool when I reached across the bench. I meant only to turn it so he could see the edge more clearly. He held it still and looked at my hand. "Are you helping, or are you taking over?" he asked. I withdrew my hand too quickly and caught my sleeve on a corner of the bench.
"I could see it wasn't seated," I said.
"Then tell me what you see." I described the gap. He turned the piece himself, checked it, and set it down. There was no quarrel worth having over the part, but my face had grown hot. I told him about Della and the missing attachment, including my offer to finish the task for her. He gave a short laugh. "At least you're consistent."
I wanted to explain that being useful had become a habit long before we planned the venture. Instead I asked him to show me how he preferred to hold the piece. He did, and when my turn came he kept his hands on his own side of the bench. By the end of the afternoon we had finished less than I had hoped. We had also worked without one of us becoming the other's assistant by default. Before leaving, I marked another completed trial on my handoff list. November was still ahead, but it was no longer merely a date I had asked other people to believe.