About a year ago, I demonstrated a method for the determination of fat in milk, adapted to human milk, and for general office purposes, before this society. Those who saw that demonstration may remember the difficulties experienced with the Babcock bottles then available. Indeed I doubt very much whether any one was stimulated to repeat the experiment. This fact has led me to devise a bottle which seeks to do away with the chief difficulties attendant on the use of those bottles. The regulation Babcock bottles may be used for office work, of course, but they involve the purchase of a special centrifuge large enough to hold them, and they call for an amount of milk (17.6 c.c.), which is sometimes difficult to obtain in the case of human milk. On the other hand, the bottles heretofore devised for office work are objectionable for one of two reasons. One form