1096 



sometimes the influence of the North-Atlantic, in other years that 

 of the Enro-Asiatic action-centre prevails, it is not « //mn' improb- 

 able tiiat tliis correlation would be more strongly reflected in an 

 area with a so unstable atmospheric eqnilibrum. An approximately 

 89-year j)eriod, which in my former investigation came out most 

 clearly of all, was also fonnd in ihe [)resent investigation or was at 

 any rate suggested by it. Now there remained one means to try the 

 reality of such a period. If it really exists, that part of the last 

 period comprised between the middle of the I9th century and the 

 present day, (the internal of about 65 years for which the theimo- 

 raetric observations in Wesleii» Europe may be considered fairly 

 homogeneous and trustworthy) must fit in with the curve 

 derived from the historical data since 760. 



Eor this test the historical as well as the modern observational 

 material had to be brought into a mutually comparable form. 



The historical records between 760 and 1916 com[)rise 13 periods 

 of 89 years. Foi' each year of the period the "temperature-coefficients" 

 were now combined, half weight being assigned to the first seven 

 periods, before 1383. These coefficients were however at the same 

 time reduced to temperature variations in degrees centigrade; this 

 proved feasible by compaiing th'e "historic" series with the observ- 

 ational series Zwanenburg — Utrecht (de Bilt) from 1734 to the present 

 day, the necessary rediu'tions having been duly applied. Although 

 such a transformation cannot of course give more than a rough 

 estimate for the real temperature variations, still for applyirig the 

 test before mentioned these values seemed preferable to using the 

 "temperature-coefficients". So we obtain, assuming a 89-year period, 

 an average of thirteen "historical" tem|)erature curves. 



Modern series of o I) s e r v a t i o n s. The period 1 852 — 

 1916 contains the 25^'' to 89^'' year of the latest 89-year period, 

 1828 — 1916. As good representatives of the Western European 

 climatic district the stations Paris (St. Maur) and Utrecht— de Bilt 

 were chosen^); for meteorological elements I took: 1. the monthly 

 averages over the meteorological winter; 2. the three minimal 

 temperatures observed in each winter but in different months '). In 



') My sincere thanks are due to Prof Angot of Paris and Prof, van Ever- 

 DiNGEN and Mr. Galle at De Bilt for communicating unpublished observational 

 data and for hints concerning the reduction of the Zwanenburg series. 



2) "Meteorological winter" always means the months December, January and 

 February taken together. Tlie winter is named after the year in which the latter 

 two months fall. Often November and March are taken into account with )ialf 

 weight, the result is then reduced to 3 months, however. 



