Toward the end of Jane Smiley's novel A Thousand Acres, the main character, Ginny Smith, goes through a period of obsessive cleaning. After a meal she washes the dishes, sweeps the floor, cleans the counters, scrubs the burners, and wipes the oven door with Windex. Then she cleans the seams in the counters and the corners and feet of the stove. With a toothpick.

We put every story through an accuracy check by a researcher.

Smiley uses the scene to show that Ginny is going slightly insane, so I regret to say how much it reminds me of the finishing stages of putting out a magazine. Publishing a factually accurate, stylistically coherent, typo-free issue does require a degree of slightly insane compulsion from everyone on staff. (You have to be not just detail-oriented but detail-infatuated to want to make sure that "West Coast," "western Montana," and "Western civilization" are correctly upper- and lower-cased throughout 136 pages.) Most critical is the fact checking. We put every story through an accuracy check by a researcher, not because we distrust our writers -- on the contrary: we are proud to be publishing some of the best writers working today -- but because we know that a scientist might not clarify to a reporter which fluoroquinones are considered second-line TB drugs, and that a dean might not mention that a student who left Yale after a plagiarism accusation hadn't been formally found guilty.

(An aside here on the Alumni Notes -- the personal alumni news we print for group and individual subscribers. We have never fact-checked the Notes. But recently, a prankster sent in a fake note about a classmate and, when questioned by his class secretary, supplied a fake confirmation from a fake e-mail address. Our faith in humankind rattled, we now spot-check the Notes.)

Fact checking is rather like law: highly detailed and literal, yet nuanced and intellectually challenging. We are extremely fortunate to have a lawyer, Thea Bourke Martin, as our principal researcher. She verifies names, titles, places, and dates, as well as scientific, legal, historic, and all other data and assertions, with both written and living sources. We discuss any factual discrepancies with the writers, and the copy is amended to be both correct in the details and true in the impression it conveys.

We still make mistakes. In our July issue, we mixed up the two 2007 graduates pictured at right. How did we manage it? We suspect misfiling. But the real problem was that I asked for a check of the graduates' interviews -- but not their photos. Mea culpa. Next year, we'll get out the toothpicks.