ASTRONOMY: C. E. ST. JOHN 
227 
portance should be kept and it becomes necessary to discriminate be- 
tween that which is possible to retain and that which must pass. 
Several attempts have been made to determine in some general way 
the systematic differences between the Rowland wave-lengths and later 
determinations based upon the international system. Fabry and Perot^ 
plotted the ratio (XR)/(X international). To reduce the Rowland wave- 
lengths to the new standard, they proposed to divide the Rowland 
values by the corresponding ratios read from the curve. Hartmann^ 
suggested the reverse procedure, namely, that wave-lengths expressed 
in the new system be multipHed by these ratios, reducing them to the 
approximate Rowland values and thus minimizing the break with the 
historical standard. Instead of the ratios suggested by Fabry and 
Perot, Albrecht in a recent paper^ uses as ordinates the quantities, 
Rowland minus international, and from the mean curve derives directly 
the systematic difference for any line. 
In many present-day discussions of solar problems, the third deci- 
mal place, when wave-lengths are expressed in angstroms, is drawn 
upon for the requisite data. One purpose of the present communica- 
tion is to direct attention to some observations which show that a gen- 
eral transformation from one system to the other is a matter of the 
greatest difficulty if the required degree of precision is to be reached, 
even though the relative wave-lengths in each system were free from 
error, and that statistical comparison between different systems is a 
procedure fraught with the possibiHties of introducing residuals that 
may be quite misleading. 
For several years direct comparison between the lines of the iron 
arc and the solar spectrum have been in progress at this Observatory, 
and recently more precise determinations of the wave-lengths of iron 
lines dependent upon arc conditions have been made. The sun-arc 
comparisons show that for the lines of the same pressure group the 
displacements between sun and arc are, among other relations, a func- 
tion of the line intensity. ^ The differences, then, between the Rowland 
and the international wave-lengths must be unequal even for lines in 
the same group and spectral region. In Table I are shown the data 
for two strong hues situated between weaker lines of the same group. 
The larger sun-arc displacements parallehng the larger Rowland-Inter- 
national differences for the strong lines are typical and the sun-arc data 
accord with the mean from a long series of determinations. 
Similar systematic differences appear when Hues of different groups, 
but in the same region, are considered, as shown in Tables II and III.^ 
