Pyrolyses of Feist's ester (IIa, [α]25D -119°) were carried to roughly 25% and 50% completion. The chromatographically separated products were Ia ([α]25D +28', +10°), Ib4 ([α]25D +18', +5°), and IIa (-93°, -34°), respectively. No cis-Feist's ester (IIb) was detected. The isolation of active product provides convincing evidence for the "valence-tautomerization" pathway (b), but the observed racemization of the starting material requires additional comment. Two racemization mechanisms not involving the "zwitterion" III were considered: (1) the trans-ester (IIa) might equilibrate rapidly with product that has partially racemized either by equilibration with the internally-compensated cis-ester (IIb) or by some other unspecified route; (2) the trans-ester might equilibrate directly with its cis-isomer. The first alternative is untenable because the experimental conditions effectively inhibit the reconversion of product to IIa or IIb due to the demonstrably favorable equilibria between the product and these esters. The second was tested by partial pyrolysis of the cis-ester (IIb) using conditions under which the trans-isomer was substantially unaffected. The resulting mixture was shown by NMR to be pyrolysis product (I), trans-ester (IIa), and cis-ester (IIb) in the approximate ratios 75:10:15. Thus, the cis-ester (IIb) rearranges about 7.5 times faster than it isomerizes to the trans-ester (IIa), and contrary to fact, racemization of IIa via IIb could occur only less than 1/7.5 times the rate of product appearance. It is thus concluded that racemization must occur by reversible equilibration of the trans-ester with the "zwitterion" IIIa (and/or IIIb) and that rearrangement proceeds competitively by both mechanisms a and b. Prodigiosin, the blood-red pigment of Serratia marcescens assigned the tripyrrylmethene structure I by Wrede, has attracted recent interest because of its possible relationship to the tripyrrylmethane type of intermediate in porphyrin biosynthesis. New evidence is presented that makes structure I untenable and shows prodigiosin is II or III. Wrede isolated 2-methyl-3-amylpyrrole (C10H17N, A) from prodigiosin, while Santer and Vogel recently showed that a mutant strain (9-3-3) of Serratia accumulated a C10H10O2N2 precursor (B) that could be converted to prodigiosin by another mutant (W-1). In a formal way, the condensation of these two C-10 fragments leads to prodigiosin (C10H17N + C10H10O2N2 → C20H25ON3 + H2O). Pure A and B were shown to react readily under conditions of dipyrrylmethene synthesis to form prodigiosin indistinguishable (by infrared spectra of zinc salt and hydrochloride) from the natural material. The NMR spectrum of B (60 Mc in dimethyl sulfoxide) contains two broad peaks at τ = -1.35 and -1.52 (two pyrrolic NH protons) and a sharp singlet at τ = 0.57 (aldehyde hydrogen). The α,α'-linkage of the pyrrole rings is shown by the isolation of pyrrolecarboxamide from the alkaline-peroxide oxidation of both B and prodigiosin. The above evidence, along with other properties of B (one OCH3, ultraviolet absorption at λmax22° 254 mμ (ε 13000) and 363 mμ (ε 33000)), supports structure II. Further refinement: (i) the strong ultraviolet absorption requires conjugation of both pyrrole rings with the formyl group, excluding any structure with -CHO at C-3; (ii) the methoxyl must be at C-3 or C-4 (based on the isolation of methoxy maleimide from prodigiosin); (iii) the NMR doublet at τ = 3.69 (which changes to a singlet in base) corresponds to the lone C-H proton on ring A; if this proton were located at C-3 (adjacent to CHO at C-2), it would be expected to show resonance closer to τ = 2.7-2.8 (as in 2-pyrrolealdehyde), so B contains OCH3 at C-3 and CHO at C-2 or C-4. While attempting to rigorously distinguish between the two possibilities for B, the α,α'-dipyrrylmethene III (formed from A + B) is considered to better accommodate the properties of prodigiosin than the α,β'-alternative II.