this location would be especially fa\ oiable. owing to the continiietl presence of 

 one or more observers. 



The frequent occinrence of extensive an<l disastrous conflagrations in the 

 City of Tokio has attracted the attention of all who have resided fur any length 

 of time in the capital. The immense destruction of property in this way within 

 a year or two has excited great interest in all investigations pertaining to the 

 origin of fires and possible means of preventing their spread. It has long been 

 known that disastrous (ires were most frequent, in general, in certain mouths of 

 the year and it is easy to see that their occurrence is intimately related to the 

 velocity and direction of the wind. Mr. Yamagawa, Adjunct Professor of I'hy- 

 sics in the Univci^sity, has recently devoted much time to an investigation of the 

 origin and course of these fiies, and to their Classification in reference to 

 atmospheric movement. The results of this investigation cannot foil to possess 

 great interest and I am indebted to Professor Yamagawa for the preparation of 

 some of the most important of these results, which will be found, with charts 

 illustrating the conclusions reached, at the end of this volume. 



BAROiMETKICAL OBSERVATIONS. 



Table A exhibits the readings of the barometer at all of the regular obser- 

 vations during the year, after having been reduced to the freezing point and to 

 the level of the sea. The actual fluctuations in the barometer during the year 

 are graphically exhibited on the six pages of Chart No I. A resume of the 

 observations contained in table A will be found in table U. In this will be 

 found the means for each hour of observation for each month in the year, as 

 well as general means, maxima and minima and range. Concerning the first it 

 will be seen that the mean for the 2 p. m. readings is the least, and that, with 

 few exceptions, the mean of the 7 a. m. is greater than those of 10 j). m. In the 

 means for the year, that for the reading at 7 a. m. is .005 inches higher than 

 that for the readings at 10 p. m. But a glance at either the table or chart, 

 given under the head of '• hourly observations," representing the means (or each 

 hour of the day, during the four months in which hourly observations were male, 

 will show, of course, that a maxiiuum docs not occur at 7 a. m. but about three 

 hoiu's later, and that 2 p. m. is not, in general, a miiiimmn point but that 

 10 p. m. represents very closely a second maximum. It is believed, therefore, 

 that the hours of observation establislied for the year 1881 will furnish a nuich 

 more truthful record of the barometrical changes. 



A brief comparison of the principal facts found in table V> and those from 

 the corresponding table in the report for 1879 will be of interest. 



