the sun than would otherwise have been the case. He thus selected the best possible opportunity. To guard against any subsequent bias on the part of the examiner of the plates, after the spectroscope had taken a plate it was then reversed, and the process repeated on another one, the iron being sparked as before. What had been the right side of Venus with regard to the red end of the spectrum thus became the left one, and vice versa. In this manner, when the plates came to be measured for tilt, the measurer would have no indication from the spectrum itself which way the lines might be expected to tilt; he could, therefore, not be influenced either consciously or unconsciously in his decision.
Eight plates with their comparison ferric spectra were thus secured; four with the spectroscope direct, four with it reversed. They were then shuffled, their numbers hidden, and given to Dr. Slipher to measure. The spectral lines told their own story, and without prompting. All the plates agreed within the margin
of error accordant with their possible precision, a precision some thirty times that of Belopolski's experiment on the same lines,—a result not derogatory of that
