From alkaloid residues derived from Crotalaria goreensis Guill. et Perr., we have isolated a crystalline base, C₉H₁₃ONCl₂, the spectral properties of which are consistent only with the structure 7β-hydroxy-1-spiro(2',2'-dichlorocyclopropyl)-8β-pyrrolizidine (I) or its enantiomer. Since the major alkaloid of this species is 7β-hydroxy-1-methylene-8β-pyrrolizidine (II) and the crude residues were kept for more than one year probably in the presence of a small quantity of chloroform (residual solvent), it seems highly probable that (I) is an artefact derived from (II). The latter is a fairly strong base and could conceivably react with chloroform to form dichlorocarbene which may then add to the double bond of the same or a different molecule. Quaternary artefacts formed from alkaloids with chloroform or its bromochloromethane impurity have been reported, but it is surprising that dichlorocyclopropyl derivatives of the above type have not previously been encountered. The compound, m.p. 170℃, is basic to litmus and does not contain ionic chlorine. It is transparent in the ultraviolet, and its infra-red spectrum shows broad hydroxyl absorption around 3000 cm⁻¹ and no absorption in the 1600-1800 cm⁻¹ region. The nuclear magnetic resonance spectrum (measured at 60 MHz) shows a strong similarity in the downfield region to that of the alkaloid (II) except that the signals due to the 1-methylene group are absent. The mass spectrum confirms the empirical formula C₉H₁₃ONCl₂ and provides confirmation of the structure (I). Despite the probable origin of (I) from dichlorocarbene and (II), initial attempts to bring about such a reaction (dichlorocarbene generated by the action of potassium t-butoxide on chloroform or ethyl trichloroacetate or by thermal decomposition of sodium trichloroacetate) have not succeeded, yielding only intractable products. Chloroform and (II), heated in a sealed tube at 150℃ for 42 hours, gave only a small amount of polymer in addition to unchanged (II). Further efforts to prepare (I) are in progress.