SSummaryInHaliotis tuberculata, Patella vulgata and Patina pellucidathe young are hatched at an early trochophore stage and remain only a few days in the plankton, but inCalliostoma zizyphinumthe whole of this phase takes place within the egg membranes and the larva is plantigrade from the time of hatching.In all four genera the larval retractor muscle is derived from mesoderm cells arising from cell 4 d, before torsion begins (Smith, 1935; Crofts, 1937). The present investigation shows that during the first phase of torsion the retractor consists of six spindle‐shaped cells all attached to the shell apex on the pre‐torsional right side. The anterior attachments on that side relate to retraction of the velum and mantle. On the pre‐torsional left side the retractor cells relate to the left side of the velum and, curving in sickle fashion above the primitive gut, have attachments to the left side of the foot rudiment. The musculature of the right side of the foot does not develop until after the first half of torsion has taken place. There is, therefore, evidence in all four genera that the larvae are asymmetrical before torsion begins.Garstang's hypothesis (1929) concerning the origin of gastropod torsion was based on a probable mutation involving asymmetry in two ancestral larval muscles, a left one related to the foot and a right one to the head. From the evidence, however, it is clear that the single asymmetrically placed larval muscle of the head and left side of the foot is the ontogenetic cause of the beginning of the torsion process.Garstang's hypothesis (1929) concerning the origin of gastropod torsion was based on a probable mutation involving asymmetry in two ancestral larval muscles, a left one related to the foot and a right one to the head. From the evidence, however, it is clear that the single asymmetrically placed larval muscle of the head and left side of the foot is the ontogenetic cause of the beginning of the torsion process.In the four genera investigated torsion takes place in two phases. The first part (90 degrees inHaliotis, PatellaandPatinaand rather more inCalliostoma)takes place with rapidity varying from three to six hours inHaliotisandCalliostomato ten to fifteen hours inPatella.The amount of yolk provided in the egg influences the length of time taken before torsion begins and therefore it affects the stage of development reached by the muscles; consequently both factors influence the time taken to complete the first phase of torsion. The position of the shell attachment of the larval retractor also affects the speed and extent of the first part of torsion. The second phase of rotation is slower, taking from thirty‐two hours inCalliostomato some 200 hours inHaliotis; it is due mainly to differential growth. Retraction into the shell is possible before 180 degrees torsion is complete.The pedal musculature of the definitive right side develops in all four genera during the second phase of torsion and assists the progress of torsion. It develops from the mesoderm of the post‐torsional right side and becomes the typical columellar muscle ofCalliostomaand the hypertrophied one ofHaliotis.Contrary to previous supposition, the retractor muscle of the pre–torsional larva does not become the columellar muscle in the genera investigated. It is probable that the definitive right shell muscle is a pair to the pre–torsional asymmetrical larval retractor, delayed in development, and that the two muscles may represent the two ancestral muscles of the bellerophonts.InHaliotis, CalliostomaandPatellathe pre‐torsional larval retractor and the definitive right shell muscle are equal in size and bilaterally situated by the time the velum has disappeared and torsion is complete. In the post‐larvalCalliostomathe columellar muscle alone persists, the larval retractor being reduced and lost as in the majority of prosobranchs. InHaliotis, however, the original larval retractor muscle persists as a small shell attachment muscle near the margin of the topographical left side; it has been assumed incorrectly that this definitive left muscle is a secondarily developed one.The specialized characters individual to all four genera investigated develop at a late period, only after the two phases of torsion are complete and the larvae have lost the velum for several days inCalliostomaandPatella; inHaliotisthe larvae have been plantigrade for a month. Although complete retraction into the shell lasts only until specialization in the post‐larval stage has set in forHaliotisandPatella, the operculum remains for some days after its function is lost. In this respect they differ fromScissurella (=Incisura)which, as shown by Bourne (1910), retains the ability to retract into the shell and a reduced operculum persists in the adult.The evidence of muscle morphogenesis and of the adult topography of the muscles of primitive gastropods generally is compared with the new palaeontological evidence published by Knight (1947–1952), who discovered paired retractor muscle scars inBellerophonandSinuites.The bearing of this evidence on the phylogeny of the families of the Archaeogastropoda is discussed.