Butterfly wing color patterns consist of many color-pattern elements such as eyespots. It is believed that eyespot patterns are determined by a concentration gradient of a single morphogen species released by diffusion from the prospective eyespot focus in conjunction with multiple thresholds in signal-receiving cells. As alternatives to this single-morphogen model, more flexible multiple-morphogen model and induction model can be proposed. However, the relevance of these conceptual models to actual eyespots has not been examined systematically. Here, representative eyespots from nymphalid butterflies were analyzed morphologically to determine if they are consistent with these models. Measurement of ring widths of serial eyespots from a single wing surface showed that the proportion of each ring in an eyespot is quite different among homologous rings of serial eyespots of different sizes. In asymmetric eyespots, each ring is distorted to varying degrees. In extreme cases, only a portion of rings is expressed remotely from the focus. Similarly, there are many eyespots where only certain rings are deleted, added, or expanded. In an unusual case, the central area of an eyespot is composed of multiple “miniature eyespots,” but the overall macroscopic eyespot structure is maintained. These results indicate that each eyespot ring has independence and flexibility to a certain degree, which is less consistent with the single-morphogen model. Considering a “periodic eyespot”, which has repeats of a set of rings, damage-induced eyespots in mutants, and a scale-size distribution pattern in an eyespot, the induction model is the least incompatible with the actual eyespot diversity.