Reflexes of *r̥ in καρτερός, κράτος and Related Forms

Among the evidence for the regular outcome of *r̥ in Ionic-Attic, the root of καρτερός, κρατύς, κράτος and related forms is of crucial importance. Several formations have doublets, the most prominent ones being καρτερός ~ κρατερός, κράτος ~ κάρτος, and κράτιστος ~ κάρτιστος. In chapter 1, we have seen that the formswith ‐αρ‐ are troublesome for accounts arguing for a regular development *r̥ > ‐ρα‐ in Proto-Ionic, in view of the full grade attested inAeol. κρέτος and Ion. κρέσσων. Indeed, by a process familiar from chapter 4, forms like κράτος have secondarily introduced an analogical zero grade κρατ‐ from a related adjective, replacing the original full grade κρετ‐. This is a very important argument for viewing ‐αρ‐ as the regular reflex. However, the analysis of doublets like καρτερός ~ κρατερός is complicated by two issues thatmust be clarified beforewe draw this conclusion. One approach to suchdoublets has been to claim that κρατ‐ andκαρτ‐were freely interchangeable allomorphs.1 This is inmy view an adhoc strategy designed to save the idea of a regular development *r̥ > ‐ρα‐. For one thing, it does not explain why doublets are attested only for a subset of the attested formations. For instance, the adjective κραταιός has no by-form *καρταιός, in spite of the fact that using κραταιός in hexametrical Greek necessitated the use ofmuta cum liquida scansion, an uncommon phenomenon in Homer. Moreover, some variants appear to be limited to specific genres: for example, κρατερός never occurs in prose. Apparently, then, doubles with καρτ‐ beside κρατ‐ could be created only under certain conditions. One main goal of this chapter is to show, by a fresh etymological analysis of the evidence, in which respective formations the root shapes κρατ‐ and καρτ‐ originated and by which mechanisms they spread. This analysis is reinforced by a close consideration of synchronic derivational relations: once we take the lexical meanings of base form and derivative into account, wemay infer withmore plausibility that certain formations were analogically or even artificially created in the epic language. For instance, we will see that the epic form κάρτος is only used as an adjectival abstract meaning