Nipher — Method of Measuring Pressure on a Structure. 13 



bottom than near the top on the back side of the board. 

 This was doubtless due to the obstructing effect of the car 

 roof. 



Fifteen observations of pressure in any square were usually 

 taken at one time and the collectors were then removed to 

 another. In the tables which follow the means of these 

 observations are given. Where several determinations are 

 given for the same square, they were usually made on differ- 

 ent days. Some discrepancies appear which seem too large, 

 but all observations have been included. The tremendous 

 shocks which our improvised laboratory sometimes received 

 made it necessary to exercise constant vigilance in detecting 

 loose adjustments, and some sources of error have doubtless 

 escaped us. 



It was found that increase of pressure on the front side of 

 the board, and decrease of pressure on the back side were 

 linear functions of the total force required to hold the board 

 to the wind, as measured by the spring balance. If .F repre- 

 sent the pull of the spring balance on an arm of 3 feet, h 1 the 

 increase in the scale reading of gauge No. 3 above the datum 

 reading, 40.0 and h 2 the decrease of gauge reading of No. 4, 

 below 40.0, then for the front and rear pressures we have 

 respectively, 



ft, = A Y F 

 h 2 = A 2 F 



where A 1 and A 2 are constants A 2 being essentially negative. 

 They denote the increase or decrease of scale reading per 

 pound of pull on the spring balance. A x and A 2 are reduced 

 to vertical water column by multiplying by 0.05. This may 

 be taken as the pressure in grammes per square centimeter. 

 The correction for density of water is about half of one per 

 cent., and is slightly overcompensated by the change of level 

 in the cistern. The further factor 2.048 reduces the pres- 

 sures to lbs. per square foot. The factor for reducing A x 

 and A 2 to lbs. per square foot is therefore 0.1024. 



The values of A v and A 2 have been entered in the proper 

 squares on diagrams of the front and the back of the pressure 

 board. Such diagrams are shown in Figs. 3 and 4. The 

 board was divided into strips where the conditions were evi- 



