The view has been advanced hy Clavel j5j and Bird [2 that disperse dyes dye cellulose acetate from , an aqueous, niolecularly dispersed solution. In this ' , process, the acetate behaves as a solid solvent, and hy degrees. a, Xernst distribution ~luitihri~u11 is reached between the dye adsorbed on the fiber and the molecularly dispersed dye in solution. It ' is beheved that the dye molecule is not bound to any of . the groups of the suhstrate or is ,only very loosely bound.. ' In our research in' this field. we have observed certain phenomena which uo not accord with these views. It was found that a given dye. exhaustiondyed from an aqueous dispersion and spun-dyed from an acetone'solutiotl in the normal way, does 110t have the spnie fastness properties (Fig. 1 ) . ' ., , These findings are incompatible with the assumption of a constant change in the' position of tl~e 'clj~e molecules in the solid solution.. If this' were so/the final state of' the dye would invariably be the same, , all the more so s18cej in both cases, the substrate is penetrated and not merely ring-dyed., .The dye distribution in these two .methods of application must differ in some way. Giles '171 [ and . Weissbein and Coven ~22J have demonstrated that a , relation exists between the physical state of dyes and ,their lightfasthess. Their experiments prove th'at t , crystalline aggregates in films are faster to light. ' With these findings in mind. we searched for evidence 'of possible crystalline inclusions in the dispersion-dyed and, more particularly, in the spun-dyed -cellulose acetate fibers [ 13 j by means of electron ' microscopy and, x-ray photography with nickt,·1-filt ered copper rays and iron rays. No crystallinity ' w~as detected even in' Spin dyeings of up to 16% ' ' depth, produced with 2:1 rnetal complex ' held 'view may he stated, which is that. in thr ,,' & d q u o ; amorphous, rubber-like fiber areas having chain iltrr . spaces of approximately ,5'A. crystallites are cm bedded which are approximately 500 A in length I approximately 50' At thick, and whose chain interval: , are about ~.5 A. By way of comparison, a (lisl)crso , dye of molecular weight 300 is a clisk with at i , average diameter of about 12 A and a thickness ti : alR>ut 4 A. In natural nbers. too, crB'~ta11ine, ele meritary l~ril5 occur which are only about 35.t liick. Viewed morphologically, it is hard to itit im . how the relatively large dye nlO~ecule~ can find ~$fli cient ...