Elliot Sober has published a claim (Sober 1999, Sober 2008: §4.1, 265ff) that Darwin used, and we should too, a particular syllogism: similarity, ergo common ancestry.
This cannot be right, for several reasons: logical, historical and inferential. First the logical, as this is rather vapid, and can be guarded against (although Sober does not so guard) relatively simply: it cannot be that similarity in itself is evidence of common ancestry, or every dice would have a common ancestor, and every rock that resembles Abraham Lincoln’s profile would too. Now the way to guard this might be to assert that yes, they do have common ancestors, in the general sense they have common etiologies. All dice resemble each other because there is a chain of cultural descent that links back to some “dice taxon” in the past somewhere in Asia. The rocks have a shared etiology in the physiognomy of Abraham Lincoln. But that is not quite the claim Sober is proposing. For this would involve the cognitive and cultural dispositions of ourselves as classifiers, and common ancestry in no way relies upon us, although our recognition of it of course does. Can we infer from similarity that the two objects that are similar (to us) have a shared causal history? The Lincoln case suggests not. One rock might be formed by a lava flow, while another might be half a world away and formed from the erosion of sandstone. Without limitations on the kind of similarity, it implies nothing at all about the objects (and perhaps quite a lot about the observers engaging in pareidola).
The historical objection is that Sober, and most other modern commentators, read Darwin wrongly. Darwin used not similarity, but affinity, as evidence for common ancestry, and technically, he inferred common ancestry from “group subordinate to group” taxonomy; that is to say, he explained this taxonomic arrangement with common ancestry, rather than defended the claim of common ancestry that way. Had he wanted to use similarity, there was a perfectly good term, before Owen’s invention of the notion of homology: analogy, as can be found in the discussions in the Quinarian literature. Darwin wrote, in chapter XIII of the first edition of the Origin:
… all organic beings are found to resemble each other in descending degrees, so that they can be classed in groups under groups. This classification is evidently not arbitrary like the grouping of the stars in constellations. 
Thus, the grand fact in natural history of the subordination of group under group, which, from its familiarity, does not always sufficiently strike us, is in my judgment fully explained. 
And he goes on to note
Naturalists try to arrange the species, genera, and families in each class, on what is called the Natural System. But what is meant by this system? Some authors look at it merely as a scheme for arranging together those living objects which are most alike, and for separating those which are most unlike; or as an artificial means for enunciating, as briefly as possible, general propositions,—that is, by one sentence to give the characters common, for instance, to all mammals, by another those common to all carnivora, by another those common to the dog-genus, and then by adding a single sentence, a full description is given of each kind of dog. The ingenuity and utility of this system are indisputable. But many naturalists think that something more is meant by the Natural System; they believe that it reveals the plan of the Creator; but unless it be specified whether order in time or space, or what else is meant by the plan of the Creator, it seems to me that nothing is thus added to our knowledge. Such expressions as that famous one of Linnæus, and which we often meet with in a more or less concealed form, that the characters do not make the genus, but that the genus gives the characters, seem to imply that something more is included in our classification, than mere resemblance. I believe that something more is included; and that propinquity of descent,—the only known cause of the similarity of organic beings,—is the bond, hidden as it is by various degrees of modification, which is partially revealed to us by our classifications. [413f, emphasis added]
Darwin goes on to discuss how external resemblances are not evidence for propinquity (nearness, or kinship). He discusses how similarity is mere “adaptive or analogical characters” and that it is “a general rule, that the less any part of the organisation is concerned with special habits, the more important it becomes for classification”. Darwin knew well about convergence. “We must not, therefore, in classifying, trust to resemblances in parts of the organisation”, he concludes. That we need an ensemble of characters, and that they are not necessarily about similarity, is clear from this passage:
The importance, for classification, of trifling characters, mainly depends on their being correlated with several other characters of more or less importance. The value indeed of an aggregate of characters is very evident in natural history. Hence, as has often been remarked, a species may depart from its allies in several characters, both of high physiological importance and of almost universal prevalence, and yet leave us in no doubt where it should be ranked. Hence, also, it has been found, that a classification founded on any single character, however important that may be, has always failed; for no part of the organisation is universally constant. The importance of an aggregate of characters, even when none are important, alone explains, I think, that saying of Linnæus, that the characters do not give the genus, but the genus gives the characters; for this saying seems founded on an appreciation of many trifling points of resemblance, too slight to be defined. 
And he then discusses affinities by saying “Our classifications are often plainly influenced by chains of affinities” . Affinities, not analogies (and as we argued, “affinity” means roughly shared sets of homologies). He summarizes by noting that
All the foregoing rules and aids and difficulties in classification are explained, if I do not greatly deceive myself, on the view that the natural system is founded on descent with modification; that the characters which naturalists consider as showing true affinity between any two or more species, are those which have been inherited from a common parent, and, in so far, all true classification is genealogical; that community of descent is the hidden bond which naturalists have been unconsciously seeking, and not some unknown plan of creation, or the enunciation of general propositions, and the mere putting together and separating objects more or less alike. [420, emphasis added]
It is plain that Darwin held that what was evidence for common ancestry was shared sets of homological relations independently of adaptive characters, which can converge. Affinities are evidence, not analogies, and Darwin knew this well.
This brings us to the inferential objection. Sober fails to deal with convergent evolution as a cause of similarity, and yet this is so well known to systematists as to be hardly worth discussing. Because he adopts what is basically a statistical notion of classification, Sober thinks, we suppose, that homoplasy, that is to say, convergence, is eliminated somehow by technique or methodological algorithms. However, every systematist strives to eliminate homoplasy before analyzing data, just as Darwin said. There is no magic method for doing this: what looks homological may turn out, upon comparison of many taxa, to be homoplasious or indeterminate, and vice versa. But despite our limitations here, we can do this successfully in most cases – if we could not, then we could not do natural classification at all.
In neither place where Sober advances modus Darwin, does he defend against this obvious objection. In conflating similarity with affinity, we are confused about what counts as evidence for a given scenario of common ancestry. Although we have suggested that there is no fixed or privileged direction of inference in a field, it does appear that if you begin with uncertainty, then recognition of naive classification based on homological relations is going to constrain and set up the explanandum for the hypothetical account to explain. The hypothesis, a historical narrative, is not evidence for itself.
Darwin is often used as a mythological figure upon whom the preferred philosophies of the writer may be painted. In that respect he is like the Bible, except that he is a lot clearer as to his intent. The actual inferential process Darwin used – the real modus darvinii[i] – is more like this: affinity, explained by common ancestry. Since affinities are groups of homological relations we might use a term of Hennig’s and say that synapomorphies give the pattern that the historical process explains. The two are not identical.
i. I am indebted to Reed Cartwright for helping me with the Latin here.
Sober, Elliott. 1999. Modus Darwin. Biology and Philosophy 14 (2):253-278.
Sober, Elliott. 2008. Evidence and evolution: the logic behind the science. Cambridge, UK; New York: Cambridge University Press.