In recent years attempts, have been made in Europe and North America to improve seed laboratory machines used for the separation of chaff from heavy seeds in grasses. The most important European contribution is probably the Gilchrist seed separator" used at the seed testing station in Edinburgh, Scotland. The Holland (Leendertz) separator has some merit, but its chief weakness as shown by Porter (7) was the inconstant speed of the motor which made it. impossible to deliver the same volume of air through the separator tube each time the valve was opened to a given point. Replacement of the motor by one that operated at constant speed was found by Brown and Porter (1) to correct the defect reasonably well. The use of bolting cloth in place of a metal screen is also a disadvantage. Following the initial work Porter (8) described a new seed separator equipped with a combined motor' and fan unit of standard specifications which made uniform separations of bluegrass, redtop and orchard grass samples either by repeated blowings of the same sub-sample or by a single blowing for each of many replicate sub-samples. He also employed synthetic samples to check the accuracy and uniformity of separation. These results were obtained with samples from the same lot and indicated that for any given lot of seed it was possible to obtain uniform results if care was used in reading the dial opening on the valve or the manometer which indicated pressure in the compression chamber. It is noteworthy that in his experiments Porter (8) showed that if the weight of sample varied from 112 to 2 grams of blue-' grass seed, the dial reading was more dependable than pressure as measured in the compression chamber, and further that pressure as measured in the blowing tube was about as reliable as the dial reading. He showed, however, that with samples having a low weight per bushel the amount of germinable seed removed at a given opening was greater than with samples having a high weight per bushel.