METEOROLOGY. 699 



to believe that throuo-h all this seeming irregularity there runs a web 

 of harmony and rhytlun. 



''The yearly aud (laily weather periods are well knowif, but these, I believe, are 

 not the only regular periods to be found. A year ago I described two rhythmic 

 periods in the weather, one of about 7\ and the other of about 6^ days in length. 

 To these I will now add two shorter periods, one of 5 days 10.8 hours, and the 

 other of about 4 days 15 hours. The action of these on the atmosphere is usually in 

 combination, but they independently vary in intensity, so that now one and now 

 the other predominates and gives its especial characteristics to the weather 

 changes." 



. Investigations of these periods in all parts of the world indicate 

 unmistakably that they exist in every part both of the northern and 

 southern hemispheres. 



Observations on dates of maximum temperatures during May aud 

 June, 1894:, are reported which show a close agreement of the observed 

 dates with those indicated by the 5J and 7^ day periods. 



''During the summer I computed a series of normal waves for each of the periods, 

 aud by taking the times of maxima and minima and the amplitude from observation, 

 I found it possible to reproduce the observed temperature with great accuracy, and 

 even to plot a curve for a week in advance which would represent the observed 

 temperature or pressure very closely as long as the periods continued to run without 

 a break. . . . The predicted and observed temperature followed almost exactly the 

 same course; but what is most remarkable is the fact that the predicted miniunim 

 temperature aud the observed minimuni temperature was the lowest observed in 

 August for more than 9 years. 



"After I had made some j)rogress in the investigation of these weather periods, an 

 experiment was begun to ascertain how far they could be made useful in forecasting. 

 Beginning with January 27, forecasts were published each week in a local newspaper. 

 These forecasts consisted in stating which days were likely to be warmest and coldest, 

 and on which rain or snow was most likely to fall. After 2 months these were 

 verified, aud it was found that of 17 days, mentioned as days on which rain was prob- 

 able, rain fell on 12 ; while on the average for each 17 days not mentioned, rain fell on 

 5, indicating that the percentage of rain was twice as great on days when rain w%as 

 forecasted as on other days. Two days were named each week as likely to be the 

 warmest days. In 6 weeks out of 8 the warmest da.y of the week occurred on one 

 of these. The coldest days were predicted in a similar manner, and in 4 cases out of 

 8 the coldest day of the week occurred on one of the days predicted. Since 2 days 

 were included in each week, the chances were that 2 out of every 7, or 28 per cent 

 might prove accidentally correct, while in reality 75 per cent of the warmest days 

 were correctly jired-icted, and .50 per cent of the coldest. 



"These facts appeared to indicate that the forecasts might prove of benefit to 

 farmers and others, and accordingly on June 2 the issue of a weekly bulletin for 

 general distribution Avas begun. These bulletins met witli an encouraging support, 

 and it was decided to continue them thi'ough the year. At the end of September, a 

 verification of the forecasts was made, aud it was found that out of 20 warm inter- 

 vals predicted 13 averaged warmer than the mean of the week, and out of 19 cool 

 intervals predicted 12 averaged cooler than the mean of the week, or about two- 

 thirds in each case. A measurable quantity of rain fell on 33 per cent of the days 

 on which rain was predicted, and on 20 per cent of the days on which no rain was 

 predicted. Reports sent me from various stations indicate that this held substan- 

 tially true for all of southern New England. The forecasts were made on Friday, 

 aud the success for the 5 days from Monday to Friday was about the same as for the 

 other days. 



