0 Orthorhombic crystals were immersed in 4 M phosphate buffer, pH 5.6. Monoclinic crystals were immersed in saturated ammonium sulfate solution buffered to pH 4.8 with phosphate. b The Euler angles (a, ß, y) position the heme relative to the crystallographic axes. c Calculated assuming the value136= 28±2. d Only two solutions for< 44.2. after projectiononto three independent directions. 16 Further-more, it is unlikely that\would lie within 5 of the projected direction of the heme normal (listed as a in Table 1) for both {001} measurements if the angular displacement of the CO bond from the heme normal were as large as13b\= 28. As a critical test for consistency with thephotoselection measurements, we use the orientationof the heme known from the X-ray structures123· 17 to calculate all values of the azimuthal angle\in the molecular coordinate system123 that are consistent with our experimental values for tan2, assuming that the polar angle\= 28±2 derived from the 10 K photoselection measurements13b is correct. With this constraint on d¡, we fail to find consistent values of\among the single-crystal measure-ments, as shown in the lower rows of Table 1. For monoclinic crystals, solutions are found only in the first and third quadrants of the molecular coordinate system for the {001} measurements, while the {100} results lead to solutionsin the second and fourth quadrants. This situation is not improved by taking the value133\= 20 found from room temperature photoselection measure-ments. Similar problems occur when analyzing the results for

Ao and A3. These results strongly suggest that structural changes of the protein among the Aq, Ah and A3 states have little effect on the average COorientation relative tothe heme. Although structural differences between crystal and solution could in principle account for discrepancies with photoselectionresults, we note that the band frequencies are not perturbed by crystallization. Thus, the minimal differences in CO orientation observed here suggest that frequency shifts among conformational substates of Mb are not primarily determined by heme-CO distortion, but are probably more sensitive to changes in distal pocket polarity. 18· 19 Further-more, the present results on P2i and P2i2i2t crystals are consistent