Discussion of Some Errors in Quantitative Spectrophotometry. ∗

Abstract

Recent advances in photoelectric spectrophotometry reported by Hogness, et al., 1 and Smith 2 now permit precise measurements of absorption coefficients. The techniques described by these workers have made it possible for investigators in different laboratories to obtain concordant results. Kuhn, 3 Miller, 4 , 5 and Van Der Hulst 6 have pointed out several analytical applications of spectrophotometry in biological chemistry. However, these analytical procedures are valid only when the investigator is cognizant of the limitations imposed by certain errors. This paper concerns anomalous transmission values (or i. e., I_o/I)† that are caused by scattered radiation or absorption due to impurities.

Data presented in Graph 1 illustrate the discrepancies in carote-noid absorption curves of solutions containing 0.6% butter and solutions with equivalent carotenoid concentrations but no butter. In this experiment 20% diethyl ether and 80% ethanol (by volume) were employed as solvent. When the absorption curves were measured on aliquot portions of butter—carotenoid solutions at one hour intervals, the difference between the last curve, in the series, and the standard curve (no butter present) was always less than when similar comparisons were made with curves obtained on earlier runs. When the absorption curves were run on butter solutions that had stood for 15 to 24 hours, no deviation was observed from the standard curves.

In experiment 2, the transmittance was measured on butter-carotenoid solutions in spectral regions where the carotenoids do not absorb. The Io deflections were adjusted to 100.0 ± 0.5 cm. Table I summarizes the I deflections for butter-carotenoid solutions at wave lengths 5600 and 6000Å. The deflections (I) for the control cell are recorded in the last column.

Data presented in Table I show that turbidity caused sufficient scattering of light to account for the anomalous absorption coefficients presented in Graph 1.

Hogness, et al. 8 have published the molecular absorption coefficients for ferriheme cyanide (Table III).

Although the data in Tables II and III are for different kinds of substances, the variations of the percent error are in good agreement. From the view point of practical spectrophotometry, it is important to recognize that a 1 % impurity on weight basis may introduce a 10% error in absorption coefficients, especially if the coefficient is very small.