EARTH'S CRUST WASHINGTON. 281 



eluded from the data selected for our purpose, for which only an- 

 alyses of fresh and unaltered rocks are considered. 



These, and other considerations that might be mentioned, tend to 

 mimimize the rather prevalent idea that the averages, such as have 

 been calculated in former years by Dr. Clarke and me, are not 

 strictly representative, in that the well-known apparent preponder- 

 ance of granitic rocks is not sufficiently emphasized. Attempts have 

 been made by some to correct such errors by weighting the average 

 analyses of the various rock types by their areal values. 11 Such a 

 procedure, however, is open to two objections: As much weight is 

 thus allowed for lava flows, of manifestly small vertical extension, 

 as for massive intrusive, bodies presumably of much greater depth ; 

 and, as Clarke points out, "the surface exposure of a rock is no 

 certain measure of its real volume and mass, for it may be merely 

 the peak or crest of a large formation." 



But the serious objection to any such attempts at correcting what 

 may be, and often admittedly are, defects in our data, is that they 

 introduce unduly the personal equation, and thus may, or are likely 

 to, introduce other errors of unknown and indeterminate magnitude. 

 As has been shown very briefly above, we are as yet in great ignorance 

 as to the igneous rocks of a large portion of the earth's surface and 

 crust, and it would seem to be the philosophical attitude to admit this 

 and, as Dr. Clarke 12 says, "do the best we can with the available 

 data." They are admittedly not ideal, but an attempt to better 

 them, at this stage of our knowledge, is more likely than not to make 

 a "bad matter worse." Let us be philosophical Italians for a mo- 

 ment, and say with them, " Ci' vuol pazienza." 



Apart from such fundamental considerations of the character of 

 our basal data as have been all too briefly touched on above, we meet 

 with others when we come to consider the analyses themselves. No 

 analyses are ideally perfect, either as to accuracy or completeness, 

 but. while it is obviously the desirable procedure to exclude from our 

 data rock analyses that may not be up to the ideal mark that we 

 may set, yet, by so doing, we shall inevitably reduce the number of 

 our data so as probably to more than offset their excellence in quality. 

 We should have and use, of course, only analyses that are perfectly 

 accurate and complete as to the determination of all the constituents 

 that may be present. But, " humanum est errare," and so we must 

 here also " do the best we can with the available data," excluding, of 

 course, from consideration analyses that are manifestly bad. Con- 



11 Daly, R. A., Igneous Rocks and Their Origin, New York, 1914, pp. 19-46, 168-170 ; and 

 Knopf, A., Jour. Geol., xxii, p. 772, li>14. 



12 Clarke, F. W., Proc. Amer. Phil. Soc, li, p. 215, 1912. 



