32 THE DETERMINATION OF LITHIUM. 



Most of the experimental work has been done with a Bunsen burner 

 having a rather small tip, which makes the flame steadier. A porcelain 

 burner was used to eliminate the copper Imes which occasionally 

 appeared, but as in most of the work the telescopes were set so as to 

 have the sodium line just barely out of the field, no other spectra 

 came into view to interfere with the observation of the lithium line. 

 A hydrogen flame which, from some preliminar}^ trials, gave promise 

 of overcoming several difficulties encountered with other sources of 

 heat was also used. With the burner which was devised for use with 

 hydrogen and the apparatus for supplying the gas the variations in 

 the intensity of the flame over a period of several hours could not be 

 controlled as well as could be done with a Bunsen burner and ordinary 

 gas, and it was finally concluded that the advantages of the hydrogen 

 flame were not sufficient to warrant an elaborate apparatus for con- 

 trollmg and supplying the gas at fairly constant pressure over extended 

 periods of time. 



To bring the solution into the flame, platinum wires are used which 

 are formed into cylinders at the ends by winding four times about a 

 No. 10 wire. The four turns lie close together making almost a solid 

 cylinder, and drops picked up in these loops are very constant in 

 weight, some of the first loops used containing 0.010 to 0.012 gram 

 of water. The cold loop is carefully plunged mto the solution and 

 taken out with the axis of the cylinder parallel to the surface of the 

 water. The drop of water is carefully evaporated by placing it at 

 such a distance above a flame that it is vaporized without spattering. 

 The loop is then brought mto the flame, usually after being warmed a 

 little. With the apparatus as used regularly in the Bureau of Chem- 

 istry different observers usually have about the same limit for the 

 amount of lithium which will barely show the lithium line. This 

 varies from 0.000015 to 0.000025 mg per cc when no large amounts 

 of other alkalis are present. 



There is a possible chance for some slight variation in the amount of 

 solution taken up by the loop, but weighings of the amounts taken up 

 in many successive trials indicated that this is a whoUy negligible 

 source of error. The bringmg of the loop into the flame is probably 

 the chief source of error. None of the authors using a platinum wire 

 made any mention of the difficulty of accomplishing this in a uniform 

 manner. If the loop is made of moderateh^ fuie platmum wire, the 

 wire is not likely to be perfectly straight, so that if it is held in a stand 

 great care is needed to have the loop always at the same height. It 

 would be entirely feasible to arrange a mechanical holder which would 

 insure the proper placing of the loop in the flame, but it has been 

 found more convenient to have one person put the loop into the flame 

 while another makes the observation. With care and experience the 

 one handling the wire can place it in the flame with the loop always 



