Here we have a somewhat different case.
The extension of 'purple' is much larger than that of 'a crystal of Chrome Alum' (apart from these purple crystals there are many more things that are purple). This is the same as was the case with "a shark is aggressive". The difference between the two predications, however, is that 'purple' in "a crystal of Chrome Alum is purple" is always on the subject (such crystals are always purple), while 'aggressive' in "a shark is aggressive" is neither constantly present in one and the same individual subject (shark) nor present in all sharks at any one time.
Now we could conclude from all this that 'purple' is per se with respect to Chrome Alum crystals (that is, all Chrome Alum crystals are necessarily purple). And because it is a quale (that is, it belongs in the Predicament of 'Quality' ) it can only figure in predicare in quale (and thus not in predicare in quid [because, in addition to 'purple'-not-representing-here-a-complete-or-incomplete-Essence, it also does not figure as if it were such an essence, as it would have done so in the predication "purple is a color" , and would then be a predicare in quid ] ). And because of the alleged per se nature of the predication "a crystal of Chrome Alum is purple" the term 'purple' (as it figures in this predication) either belongs to the Predicable ' Difference' or to the Predicable ' Proprium'. But it can be neither of them because its extension is larger than that of the subject ( = a crystal of Chrome Alum). Recall that the extension of a difference must fully coincide with that of the species, and the same goes for a proprium, unless we consider a generic [instead of specific] proprium, but then the collection of objects possessing this proprium must be such that it can be signified by a true genus (or higher logical class term). This is, however, not the case with the domain of objects having a purple color : The collection of all Chrome Alum crystals + all other purple objects is not such that it can be signified by one or another genus (or higher logical class term). It is a collection of disparate objects, generally, having only their purple color in common.
So the predication "a crystal of Chrome Alum is purple" is accidental after all, despite the fact that such crystals persistently have this color.
In every Chrome Alum crystal there must be some physical or chemical structural aspect that is responsible for the color purple. This structural aspect absorbs certain wavelengths of incident (white) light and reflects and scatters others, resulting in the crystal looking purple. So 'being purple' can be referred back to this structural aspect, which aspect is permanently present in the crystal. And the same structural aspect, or one that has the same effect when the object having that aspect is struck by light, is present in all the other things that are purple. Can we conclude from all this that 'being purple' is accidental for the Chrome Alum crystals? All we can say for the moment is that the mentioned structural aspect is a necessary consequence of the dynamical law (crystallization law) of crystalline Chrome Alum, and so these crystals are necessarily purple. However, 'being purple' (and, correspondingly, 'possessing the mentioned structural aspect' ) is by far not a unique property exclusively occurring in Chrome Alum crystals. Only in that sense 'being purple' is accidental for Chrome Alum crystals.
Of course not only Chrome Alum crystals present this problem of the status of their intrinsic color. All intrinsically colored crystals do present it, and, in addition to them, many, if not all, other things.