An analysis of published X-ray diffraction data from nerve myelin is given based on the properties of analytic functions. Functions defined by a finite Fourier transform may be described by their distribution of zeros. This description allows a phase function to be determined from real data, which is unique in principle. A solution to the phase assignment is given and compared with corresponding published solutions derived by other methods. The strong measure of agreement for the phases of thc first nine diffraction orders, and the stability of this agreement against the efforts of experimental error, leads to the conclusion that these phases are probably correct.