[farmer] determination OF COEFFICIENT OF DISCHARGE SI 



one shoot. At the commencement of the experiment the other shoot was 

 moved over by means of a lever, so as to catch the water and guide it 

 into the measure. In the same way the water was cut off' and again 

 allowed to run to waste when the measure was nearly full. 



The chronograph has a revolving cjdinder driven by weights and 

 regulated by a governor. This cylinder carries a roll of paper on which 

 the record is traced. A record of seconds is traced by a glass pen, con- 

 sisting of a continuous line with indentations corresponding to successive 

 seconds, the motion of the pen giving these indentations being caused hy 

 an electric current being closed at every passage of the pendulum of a 

 clock in the laboratory, the current passing through an electro-magnet 

 connected with the pen. Another glass pen followed in the same line, 

 and was connected with a similar mechanism. The lever which threw 

 over the shoot conveying the water to the measure in its passage pressed 

 a button completing another circviit, which thus caused the second pen to 

 mark the exact time of start and finish of the experiment by other in- 

 dentations. In estimating the total time of an experiment the fractional 

 parts of a second were measured on the record with a rule ; the length 

 of the second-intervals being about -4 inch. 



The time of the experiments was also checked with a stop-watch, 

 but the times so obtained were found to ditfer somewhat from those given 

 by the chronograph, no doubt owing to the personal error in starting 

 and stopping. In this direction greater accuracy has therefore been 

 obtained than in previous experiments carried out in the same laboratory, 

 and probably than in those made elsewhere. 



It will be interesting to estimate the degree of accuracy with which 

 the dilïerent factors in the expression for the coefficient can be obtained. 

 Taking thejn in order, 



Q 



C = — =■ 



' At -s/'^gh 



The quantity is measured by means of a rule graduated to hun- 

 dredths of an inch, 39 inch representing a gallon. The reading would 

 probably be correct to ^j^ inch, or about -Jq- gallon. Taking the error as 

 Jq gallon in 100 gallons, gives an error of 1 in 5000, or '0002. 



The area of the orifices as measured in the comparator would be 

 liable to a little uncertainty, on account of the irregularity of the edges 

 seen under a high magnifying power. As a mean of a number of mea- 

 surements was taken, it is probable that the area would be correct to 1 

 part in 10,000, or -0001. 



The error in measuring the time might amount to ywô ii^ch or ^ 

 second. In the least time of an experiment this would be an error 

 of '00005 ; in the greatest time, 40 minutes, the error would be only 

 00001. 



L I t. 



