Passage
Temperate-zone fruit trees will not bloom in spring unless they have first spent a sufficient interval in the cold. Growers therefore need a way of saying how much cold a winter supplied, and for most of the last century they have said it in chill hours: the number of hours in which the air temperature stood between roughly zero and seven degrees Celsius. A cultivar is then described as requiring, say, nine hundred chill hours, and a site is judged suitable if its winters usually deliver that many.
The count’s convenience is considerable, and it is worth separating the sources of that convenience from the question of whether the count is sound. A total can be estimated from the daily minima and maxima that ordinary stations have recorded for a century. And because advisory bulletins, nursery catalogues and supply agreements came to be written in the same units, a grower who abandoned them would have had nothing left to compare.
None of this is evidence that hours between zero and seven are what a bud responds to, and the accumulating objections suggest they are not. Hours near the middle of the range appear to contribute more than hours near its edges, so two winters with identical totals need not be equivalent. Warm afternoons appear to cancel some of the chill already accumulated, which a running total cannot represent, since it can only rise. Models that weight hours unequally and let the accumulated quantity fall track flowering in mild regions considerably better than the plain count does.
The plain count has nevertheless not been displaced, and the reason usually offered, inertia, is only part of it. The alternatives require hourly temperature data, which older archives rarely contain. More awkwardly, a cultivar’s requirement is not established independently of the count. It is obtained by finding the total that the count assigns to winters after which the cultivar bloomed normally. Change the index and every requirement has to be determined afresh; until it is, a contest between the two indices is a contest between a calibrated instrument and an uncalibrated one. The plain count survives less because it is defensible than because it is embedded in the arithmetic of everything that has been measured with it.
