The University's Blue Hills Observatory has worked out a "table of probabilities," which predicts temperatures for fall and winter days.
Meteorologist John H. Conover of the Observatory staff has reduced the climate to a code by studying records dating back 70 years. He claims that "occasionally the critical temperature will occur on a date differing widely from that anticipated. But the probabilities are accurate and if the graphs are played each year, results. in the long run, should be good."
Conover did the work for the Monsanto Chemical Company, which needed some information on low temperature dates because it manufactures anti-freeze. His assignment concerned motorists' deadlines for installing anti-freeze. Three temperature levels were used in plotting his curves, 32 degrees, 20 degrees, and 10 degrees, for the period from September to March.
The code of probability shows that on Thanksgiving Day in Boston there is a 97 percent chance that the temperature will be 32 degrees or lower.
In Boston, the curve for 32-degree weather begins on October 11, and by November 21 there is a 100 percent probability for 32-degree-or-below weather.