28 BULLETIN 1233, U. S. DEPARTMENT OF AGRICULTURE. 



those especially covered by the present study in considering air 

 temperatures of the various forest types. 



As already stated, the air temperature observations in this study 

 fall into two groups, namely, (1) those referring to conditions several 

 feet (4J or more) above tne ground, and hence more or less repre- 

 sentative of an extensive locality; and (2) those applying to condi- 

 tions within a foot of the ground and reflecting the local influences 

 of insolation and radiation. Data of the second class will be referred 

 to as "ground temperatures" and are available for only a few of the 

 special stations where the conditions directly affecting forest repro- 

 duction seemed very important. 



By agreement the mean temperature for any period of 24 hours 

 may be computed by adding the highest and lowest temperatures in 

 that period and dividing the sum by 2. It is well understood that 

 for any single period tins method is not at all likely to give a true 

 mean temperature. But common experience shows that over a 

 number of daily periods (for example, a month) the mean tempera- 

 ture so computed will be practically the same as the mean computed 

 from hourly temperatures. The latter, of course, can be had only 

 where a thermograph is employed, and to make use of the many 

 records obtained from maximum and minimum registering ther- 

 mometers it is necessary to employ the simpler method. It seems 

 desirable, therefore, to keep all records to the same standard, even 

 where hourly temperatures might be secured from thermograph 

 traces. 



In addition to the convenience of the method the mean tempera- 

 ture derived from the maximum and minimum temperatures lias 

 this virtue — that it involves no element of judgment in the computa- 

 tion. Without going into the mechanics of the thermograph, it may 

 simply be said that the correction to be applied to different portions 

 of a thermograph t^ace can not properly be determined by rule-of- 

 thumb as is usually done. The character and amount of such 

 corrections must rest upon the question of whether or not the observ- 

 able errors in the trace are due to the natural inertia of the instru- 

 ment or to improper adjustment of its mechanism. Hence slightly 

 variable results may bo obtained in reading hourly temperatures 

 from the thermograph trace, and this fact detracts greatly from the 

 value of hourly records as ordinarily prepared. 



The temperature records of the control station at Fremont have 

 been prepared by computing the maximum and minimum tempera- 

 tures for each midnight- to-midnight period from the thermograph 

 trace as corrected from thermometer registrations. The daily range 

 is the difference between the highest and lowest temperature from 

 one midnight to the next and the daily mean the average be! ween 

 the highest and lowest temperatures. To obtain records for other 

 stations exactly comparable with the control station as to period, 

 the same method has been followed where thermograph traces made 

 it possible. For (hose stations having no thermographs the method 

 has always been bo make observations in the morning, tabulating 



the minimum temperature as of the day on which recorded and the 



maximum temperature as of the day previous. 



