110 Penfield and Foote — Composition of Tourmaline. 



fluorine is estimated the water is given as equal to the loss on 

 ignition less either the fluorine or the silicon fluoride equiva- 

 lent to it, while when no fluorine was determined a quantity is 

 assumed, usually one-third to one-half of one per cent, and 

 this quantity or the silicon fluoride equivalent to it is deducted 

 from the loss on ignition. In the latter case the assumed 

 quantity of fluorine never appears in the results as a constitu- 

 ent of the mineral. Again, no account seems to have been 

 taken of the fact that on prolonged ignition the ferrous iron 

 may become oxidized to ferric, thus causing the water deter- 

 minations to be too low. In this connection it is interesting to 

 refer to some results obtained by Biggs, who has given with 

 many of his analyses both the loss on ignition and the direct 

 water determinations. When much ferrous oxide is present 

 the loss on ignition is less than the water, but when this loss is 

 increased by an amount equivalent to the oxygen which the 

 iron present can take up in changing from ferrous to ferric 

 oxide, the results become almost equal. The results, in fact, 

 would almost do for duplicate determinations, as shown below : 



Number 36 37 38 41 42 44 45 46 47 48 49 50 51 



Increase of 0. *03 -06 -15 '35 -72 1'34 1-58 152 1-32 *94 -91 -91 -42 

 Loss on ign...4'35 410 4-09 3-62 3'3l 2-30 2*17 2*19 2*41 288 286 2'69 3-59 



4-38 4-16 4-24 3'97 4'03 364 3*75 3'71 3"73 3-82 3-77 3'60 4'01 

 H 2 direct ...4-26 390 4-16 3-64 3"49 3-69 362 3-49 3"62 363 3-60 3*34 3-79 



We cannot agree with Bammelsberg that the method adopted 

 by Biggs for determining water directly gives too high results, 

 for it has been abundantly proved that when properly executed 

 it gives exceedingly accurate ones. It is safe to assume, there- 

 fore, that if the determinations of the loss on ignition as 

 given by Rammelsberg are corrected for the oxidation of the 

 iron, the results will furnish fair estimates of the percentages 

 of water. 



Accepting as an established fact that Si0 2 and B 2 O a are 

 always present in tourmaline in the proportion 4 :1, Bammels- 

 berg's results may be modified by correcting the water deter- 

 minations and estimating both Si0 2 and B 2 3 by difference, 

 thus distributing the errors of analyses upon two constitu- 

 ents, rather than upon the B 2 3 alone. Below are given the 

 results of applying this method of correction to seven of 

 Bammelsberg's analyses made upon varieties of tourmaline 

 which have subsequently been investigated by either Biggs or 

 Jannasch and Kalb. There are also given for comparison the 

 analyses of the other investigators and Bammelsberg's figures 

 for SiO a , H 2 and loss on ignition. 



