If you have any interest in mad cow disease and any suspicion that maybe we're not being told the whole truth about it, you need to read "The Case of the Cherry Hill Cluster," from this week's New York Times Magazine. Max writes about Janet Skarbek, who's discovered that there were eight victims of Creutzfeldt-Jakob disease (CJD) among people who ate food served at Garden State Race Track in Cherry Hill, New Jersey, between 1988 and 1992 -- far, far more cases than would be expected based on CJD's prevalence in the population at large. (Mad cow disease is one form of CJD.)

D. T. Max, the author of the article, presents a lot of reasons for doubting that Skarbek is on to something -- but none of them are very convincing, and one seems flat-out wrong. He writes,

Originally [Skarbek] was interested in victims who ate at the track often. Eventually she was interested in victims who ate there even once. This increased her pool considerably. Attendance at the racetrack from 1988 to 1992 was at least four million people.

I think he's misinterpreting what he was told. I doubt the racetrack was visited by four million unique people in that four-year period. That's just now how sports attendance figures are usually reported.

When the New York Yankees, for instance, report that their 2003 attendance was 3,465,600, they mean that an average of 42,785 tickets were sold for each of their 81 home games -- but plenty of people attended more than one game in 2003, and an individual who attended, say, five games was counted five times. (A full-schedule season-ticket holder would be counted 81 times.)

Maybe Garden State Race Track carefully identified each ticket holder over that four-year period and factored out the duplicates when reporting total attendance -- but I doubt it. I assume four million tickets were sold -- to far fewer than four million people. (Horse racing doesn't attract a lot of casual fans these days -- it's mostly aficionados and, well, gambling addicts. So I think it's a safe bet that many if not most ticket buyers were repeat customers.)

So if there were eight CJD cases among two million people who went to the track in that four-year period, or one million, that's a big cluster -- in the U.S., as the article notes, you have a one-in-a-million chance of getting sporadic CJD.

A rate two, four, or eight times the national race could still be pure chance -- but it's worth looking into.