5o8 Journal of Agricultural Research voi. xvin. no. io 



The complete solution of the problem of determining the probable tem- 

 perature at some particular place, such as Denver on June 15 at 2 p. m., 

 would be to determine first the mean daily temperature for June 15 by 

 means of equation 4, using as the first term of the equation the mean 

 annual temperature for Denver and changing the date to degrees as ex- 

 plained earlier. By means of equation 6 just as it stands, change 2 p. m. 

 to degrees and insert this value for d in the equation and thus deter- 

 mine what percentage of the mean temperature the 2 p. m. temperature 

 is. Multiply this percentage by the mean daily temperature obtained 

 from the solution of equation 4 and get the desired 2 p. m. temperature. 



The lowest monthly temperature given in Table I is for January at 

 Uinta and is 16° F. With a daily variation of only 15°, a minimum of 

 8° would be expected. On certain days, however, negative tempera- 

 tures are experienced. If the mean gets too low, some correction must 

 be made, because the numerical solution does not, as it stands, allow 

 negative values. When the mean temperature gets as low as 25° or 

 lower add 20° to the mean, multiply by the percentage given in the 

 table, and then substract the 20°; the normal temperature will be 

 obtained. This will accommodate itself to negative values also. 



The methods described above (particularly the trignono metric one) for 

 determining the normal temperature give very accurate results for dry 

 and humid regions. There are, however, two sources of error. The 

 Weather Bureau record might not cover a long enough period of time, 

 and the addition of another year's values might change the normal that we 

 have used in making the equation. Also the equation, which is a series, 

 might not include enough terms to have it represent exactly the normals. 

 As explained earlier, these errors are very small. 



Actual temperatures depart more from the normal in humid regions 

 than in dry sections, and in comparing the normal as calculated by the 

 foregoing methods with a particular observed temperature, the departure 

 will be considerable in the humid section but only slight in dry regions. 

 In the western part of the United States between the Rocky and the 

 Sierra Nevada Mountains, an average yearly temperature departure exceed- 

 ing 0.5° F. is unusual. In the average monthly temperatures, a departure 

 from normal of 2° is common, but a departure of 4° is unusual. For the 

 daily temperature, a departure of 4° from the normal is common, and a 

 departure of 10° is unusual. 



The States of Idaho, Nevada, Utah, Arizona, New Mexico, Montana, 

 and Wyoming all have a relative humidity of about 50 per cent and an 

 annual precipitation of from 10 to 20 inches — that is, the precipitation 

 is light and the atmosphere comparatively clear. About 300 days out 

 of 365, or about 80 per cent, are days free from rain, with the air clear 

 and the sunshine bright most of the time. Better results, therefore, can 

 be obtained in forecasting the temperature in this section of the country 

 than in the sections where rains are frequent and the thermograph records 

 are irregular in shape (see fig. 6). 



