To illustrate: If the operator found that he walked at the rate of 43^ feet per 

 second and he wished to use the 1-64-3000 formula, a glance at the table would tell 

 him to insert in the union the disk stamped with the figure 83. Such a large num- 

 ber of disks will be needed only by the experimenter. The practical grower would 

 not have occasion to use more than two or three and could make them as desired. 

 This method of regulation was found to be very satisfactory and has the advantages 

 of (1) permanent accuracy; no variation with wear or play; (2) cheapness and ease 

 of construction within the reach of any grower; (3) simphcity; no complicated 

 mechanism to be frequently out of repair; and (4) great elasticity, since a disk may 

 be quickly made to suit any desired rate of flow. 



The next modification was made for the purpose of correcting the variation in 

 flow due to difference in head of the liquid as the quantity in the tank diminishes. 

 AU of the tanlcs which the writers have seen described have loose-fitting Uds in the 

 top through which the solution is poured to fill the tank. As the solution flows out 

 through the outlet pipe the space which it leaves in the tank is immediately filled 

 with air which comes in around the hd. Such an arrangement makes the rate of 

 flow entirely dependent on the height of the water in the tank and as a result, the 

 flow is much more rapid when the tank is full than when it is nearly empty. In 

 order to determine the extent of such variation, the 14-quart tank just described 

 was filled and the water permitted to run out with the lid loose or removed entirely. 

 The time required for delivering each of the successive quarts was determined with 

 a stop watch and the average of three tests is indicated in Table 2 : 



TABLE 2. 

 Rate of Flow of Each Successive Quart of Liquid from 14-Quart Tank (Open) 



Thus, the operator who started with a full tank would be putting on the first 

 part of the treated row double the amount which he put on the last part. The re- 

 sulting control or injury would naturally show a similar variation. Data derived 

 from the use of such a tank cannot be accurate. The variation in even the small 

 tank described by Stone has been noted above. In order to make our control data 

 reUable it was absolutely essential that this variation be eliminated. After consid- 

 erable experimentation this was finally very successfully accomplished by modifying 

 the tank thus: Instead of a loose-fitting Ud, a screw cap (B), of 2-inch diameter 

 (such as is commonly used on an automobile radiator) is used for closing the opening 

 in the top. When screwed down with the shoulder against a rubber washer, this 

 makes the top air-tight. A stand-pipe (D) parallel with and as high as the side of 

 the tank is connected with the air vent (C) by means of an elbow.* 



In this way, atmospheric pressure is maintained at the level of the vent (C) and 

 consequently a uniform flow of solution is obtained irrespective of the height of 

 water in the tank. The effective head producing the flow is always equal to (H). 

 In order to prove that the flow is uniform, the outflow of each successive quart was 

 timed just as described above but with the air-tiglit cap screwed down, and (with 

 the No. 150 disk in the union) every quart from the first to the 14th flowed out in 

 just 373^ seconds. 



*In a later model this stand-pipe is built inside the tank and is simpler in construction but the principle 

 is the same. 



22 



