This disadvantage is got rid of by making the mouth of the jet the lower limit, or, in other words, allowing the instrument to empty itself. There are two forms of such pipettes; in the one generally recommended in Gay-Lussac's silver assay (the last shown in fig. 29) the nose is replaced by a jet. This is most conveniently filled by stopping the jet with the finger, and allowing the liquid to flow in a fine stream into the neck until the pipette is filled, and then working as just described. The other form is the one in general use; in fact, a long nose to a pipette is so convenient that it may almost be said to be necessary. But the accuracy is slightly diminished; a long narrow tube makes a poor measuring instrument because of the amount of liquid it finally retains. A defect possessed by both forms is the retention of a drop of varying size in the nozzle. Whatever method is adopted for removing this drop must be always adhered to. The most convenient form is the one last described, and the most useful sizes are 100 c.c., 50 c.c., 20 c.c., 10 c.c., and 5 c.c. Ten c.c. pipettes graduated into tenths of a cubic centimetre are very useful: those are best in which the graduation stops short of the bottom. All measurements should be made at the ordinary temperature; and, before being used, the pipette should be rinsed out with a cubic centimetre or so of the solution to be measured. After using, it should be washed out with water. ~Burettes~ differ mainly from pipettes in having the flow of liquid controlled from below instead of from above. The best form is that known as Mohr's, one kind of which is provided with a glass stopcock, while the other has a piece of india-rubber tube compressed by a clip. The latter cannot be used for solutions of permanganate of potash or of iodine, or of any substance which acts on india-rubber; but in other respects there is little to choose between the two kinds. A burette delivering 100 c.c., and graduated into fifths (_i.e._, each division = 0.2 c.c.), is a very convenient size. For some kinds of work, 50 c.c. divided into tenths (_i.e._, each division = 0.1 c.c.) may be selected.