1094 



otlier. Om llie other liand it seemed absolutely necessary to restrict 

 liie critically exauiiiied matei-ial as imicli as possible to a 

 single climatic province. In the older works, e.g. Pii.gram's and 

 Schnurrer's — on which Köppen's material is largely based — areas 

 have been mixed up (Southern Germany, Austria, Italy, France etc.) 

 which occasionally must yield contradictory results. 



The remaining uncertainty in the data so obtained is chiefly caused by 

 the very vague accounts of mild winters, for which there is not, as 

 in the case of the cold ones, an incontestable criterium, e.g. ice- 

 formation, or the blocking up of traffic ; and in a lesser degree by 

 the difficulty of determining the character of those winters which 

 deviate but little from the normal standard or present an uncertain 

 character. A closer investigation might only partly obviate this uncer- 

 tainty. On the other hand, it ap[)eared in the course of this inves- 

 tigalion that the data concerning winters with an unmistakably cold 

 or mild cfiaracter, even for very remote times, are more complete 

 and reliable than might reasonably be expected, so that they con- 

 stitute a souml basis for investigation, not only because they 

 represent the only available information ()receding the thermometric 

 series, but also for their intrinsic value. 



After the historical material had been carefully collected and 

 compared 1 assigned to each abnormal winter a positive or 

 negative coefïicient of temperature, ranging from -|- 5 to — 5^), the 

 others being taken as normal and marked zero. As a starting point 

 I took the year 760, i.e. the winter 759 — 760; previous to 

 the middle of the eighth century the information seemed rather 

 doubtful and too incomplete. From the frequency of the abnormal 

 winters after the 14th century it would follow that, from that 

 time, the data, at any rate those referring to the cold winters, are 

 fairly complete, so that over the last five centuries only very few 

 abnormal winters seem to be lacking in our list. The whole series 

 760- 1916 contains 1157 years. 



This series was then submitted to an inquiry for a possible period- 

 icity, without any presupposition and without starting from solar 

 periodicities, as was done in the former investigation. Four methods 

 were applied : 



1) See Peterm. Mitt. I.e. 175. These "temperature coefficients" have been 

 assigned to the winters in a similar way as Koppen applied the values for his 

 "hervorragende W^inler". 1 adopted the scale — 5 to -f-5, however. The enumer- 

 ation of the coefficients assigned to each winter and the classification of the 

 abnormal winters according to their severity, duration and extent, must be 

 reserved for a more elaborate paper. 



