Noll RESEARCHES ON METEORITES— MERRILL. 27 



It may be added that there must invariably be more or less variation in the proportional 

 amounts of the different elements as found by analysis, owing to the difliculties in sampling, 

 without sacrificing what is felt to be too large a quantity of material, a rock in which metallic 

 iron is so prominent a constituent. This might account for the failure on the part of some to 

 detect platinum and allied elements, or such minerals as oldhamite and the problematic phos- 

 phate. An analysis of the stony forms, accompanied, as it always should be, by an examina- 

 tion in thin section, leaves, however, little excuse for lapses of this nature. The reported occur- 

 rences in the older analyses of elements not found in later investigations may very naturally be 

 ascribed, in part, to impure chemicals. 



VI. TABLE OF ANALYSES AND DISCUSSION. 



In the following table the writer has brought together the more satisfactory complete analyses 

 of stony meteorites made during the progress of this investigation, as well as such made 

 by others as seem up to the modern standard. In this work of compilation, analyses have been 

 ruled out in most cases in which the totals fell 1 per cent and more short of 100 per cent, and 

 also those that footed up approximately to 100 per cent, but in which certain elements which 

 were obviously present were not mentioned. It is, of course, possible that in all cases full 

 justice in the selections has not been done to the other workers, but the error, if there is one, is 

 of omission rather than commission. 



The purpose of tabulation in this form is to render the analj-ses comparable with those of 

 terrestrial rocks. In the work of preparation it has been necessary to recalculate in part sev- 

 eral of them. Those constituents the statement of which required most frequent attention 

 have been the ferrous sulphide which has been tabulated as Fe and S; the ferric oxide, which is 

 mainly secondary, and has been recalculated as ferrous, and the chromite which has been recal- 

 culated as chromic oxide (CrjOj) and ferrous oxide (FeO) , a proceeding which is recognized as not 

 absolutely correct, since the meteoric chromites almost invariably contain several per cent of 

 alumina and appreciable quantities of magnesia. The phosphorus has been allowed to stand as 

 given, either as P or P2O5, though the probabilities are that it belongs in aU cases to the sihcate 

 portion, and to be consistent should be tabulated as P2O5. The recalculation and tabulation, of 

 the iron in the sulphide as Fe is also open to question, and hence in such cases the percentage 

 amounts are inclosed in brackets. In all these cases the sulphide is assumed to be in the form 

 of FeS. Where an element has been recalculated as an oxide or the reverse, allowance for the 

 gain or loss in the total percentages has been made in the customary manner. 



It is to be noted that one of the most common and widely disseminated of the minor mete- 

 oric constituents — chlorine — has been ignored in this investigation, as in that of the majority 

 of workers on stony meteorites. This is due in part to the ready oxidation of the mineral law- 

 rencite, in which it mainly occurs, and in part to its seemingly trifling amount. Its presence 

 in other form of combination than with iron — and perhaps nickel — has yet to be satisfactorily 

 shown." As wiU be seen by reference to the table, I have found but five recorded determina- 

 tions from which to calculate averages. 



» Sw Cohen's Meteorit«Qkunde, pp. So aad 227. 



