Does the acquisition of words like “dog”, “table”, and “sand” require the support of sortal concepts? In arguing for and against sortals, theorists typically contrast representations of unsorted individuals (bare particulars) and sortal representations. As such, bare particular and sortal representations are presented as alternative means of representing concepts like “dog”, “table” and “tree”. Arguments for and against sortals typically proceed in the absence of an explicit characterization of the form of sortal representations. I present an explicit theory of the form of sortal representations. It turns out that, for sortals to do the work they need to do, they must incorporate bare particular representations into the sortal representation. Two experiments provide evidence for the predicted by the proposed theory of sortal representations. I also show how the proposed theory of sortal representation is consistent with recent findings by Rips and colleagues that seem to provide empirical evidence against sortals.