[ 3S9 ] 
level : but by either blowing acrofs the round hole, or 
moving his finger over it, he will foon bring the water 
in the kneed tube to fiand at the fame level with it, by 
taking off gradually the convex fuface of the water, 
which projeffs out at the hole in the form of a drop or 
fpherule. And this effeft the wind very foon produces 
itfelf. There ought always to be a cover on the top 
of the tube out of which the water is expelled by the 
wind; but it fliould be made very thin. For if there be 
no fuch cover, and the mouth of the kneed tube be ftop- 
ped, after the inffrument is quite full of w'ater, in order 
to prevent the wind from having any influence in raifing 
it, you will find, upon expofing it to a ftrong gale, that 
in a very fhort time it will blow out perhaps half an inch 
of water. Whence it appears, that a very confiderable 
error would arife from ufing the wind-gage in this ftate. 
But in all the experiments which I have made with 
this inftrument, whilft it had the cover and the round 
hole of -i^ths of an inch in diameter in the middle of 
it, I have not been able to difcover any error. The 
ufe of the fmall tube of communication a b (fig. i .) is to 
check the undulation of the water, fo that the height of 
it may be read off from the fcale with eafe and certainty. 
But it is particularly defigned, to prevent the water from 
being thrown up to a much greater or lefs altitude, than 
the true height of the column, which tho wind is able at 
that time to fuftain, from its receiving a fudden impulfe, 
whilft it is vibrating either in its afcent or defcent. For 
water in the legs of a fiphon is capable of being puti 
VoL. LXV. C c c into 
