744 
PROCEEDINGS OF SECTION II. 
The pressure on a vertical surface being 100, the presence or 
absence of the other slope of the roof CD did not perceptibly affect 
the result. 
Next the roof shown in Fig. 5 was tested. In this BC=CD = EF. 
The angle of slope is 30°, and the point E is at the same level as C. 
It was found that, the pressure on a vertical surface being 100, that on 
EF is 10 if DE is closed, and 20 if DE is left open. 
In Fig. G the upper portion of the roof EF is on the same straight 
line as BC, the angle as before being 30°. Here the pressure on EF 
was 15 when DE was closed, and 45 when DE was opeu. 
Fig. 7 represents an experiment to determine the amount of 
shelter a roof received from a precisely similar roof windward of it. 
A s might naturally be expected, the shelter was practically complete, 
the pressure on EF being too insignificant to measure. 
Figs. 8 and 0 represent experiments to determine the wind 
pressure on a veranda of ordinary proportions connected with a one 
and two storied building respectively. In Fig. 8 no pressure either 
upward or downward could be measured at a slope of either 30° or 
45°. In Fig. 9, at 20° slope, no pressure was recorded, but at 45° the 
top of the veranda experienced a pressure equal to 20 per cent, of 
that on a vertical surface. 
In Figs. 4, 5, and G a moderate projection of the eaves over the 
wall did not appreciably alter the result, but if the wall projected as 
a parapet above the junction of the roof, the pressure was always 
reduced, and with the flatter pitches sometimes rendered negative. 
The foregoing investigation confirms Dines* formula connecting 
velocity and pressure, and controverts Crosby’s and Smea ton’s. It 
gives a probably more accurate representation of the effect of varying 
angle of incidence than either Duchemin’s, Hutton’s, or Crosby’s. It 
gives the outline of the “ wind shadow” of a wall within the limits of 
which the wind pressure is practically inoperative, and it supplies data 
for computing the effect in pressure on certain commonly occurring 
arrangements of roofs and verandas, in connection with the walls of 
buildings, which, as far as I am aware, have not been previously 
studied or intelligently treated. 
It is to he admitted that the experiments have all been on a 
comparatively small scale, and that the velocity range has not been 
very large. " The apology for this is limited time and restricted 
facilities. Unquestionably it is most desirable that some person or 
body possessed of ample time and funds should repeat them on a 
tenfold larger scale, and with the utmost attainable precision. As 
Crosby says at the close of his paper, “The cost of determining 
formulae true beyond any question would he repaid many times every 
year in applying them to the true economy of engineering structures.” 
Note.— During the above experiments the barometer read approximately 30 inches, 
and the thermometer 70° F. 
