In The human penis as a semen displacement device, a 2003 paper published in Evolution and Human Behavior, a group of SUNY Albany psych researchers investigated the shape of the human penis to discover whether it could aid in "sperm competition," driving sperm out of the vagina prior to its own deposit -- something already observed in damselflies.
They devised an ingenious experiment. They produced a cornstarch mixture that approximated human semen in the judgment of "three sexually experienced males," and deposited a single ejaculation's worth of the mixture in a transparent artificial vagina. Then they plunged a penis-shaped dildo -- with a flared glans -- into the vagina and observed that the action of the penis did, in fact, force the "competing" semen behind the glans, which then scooped it out of the vagina. A smooth dildo -- used as a control -- did not scoop the "semen" out of the artificial vagina.
The authors note that their on-campus surveys found that 12 percent of the students polled had had sex with multiple men in a 24 hour period and hypothesized that an evolutionary function of the glans is to force out the semen left by earlier partners prior to ejaculation.
The human penis as a semen displacement device
In the final phase of Experiment 2, we assessed the effect of the coronal ridge on semen displacement. Using a knife, the coronal ridge was removed from phallus D taking care not to destroy the glans. Fine sandpaper and an emory board were used to smooth the area that had been cut away, and the resulting phallus (designated D ) is depicted in Fig. 5. Using the low-viscosity flour and water semen mixture, five displacement trials were conducted with both the intact and modified phallus.
3.2. Results As was true for cornstarch, the dilute semen lead to comparable displacement effects as a function of phallus morphology. Phallus B displaced an average of 97.8%, phallus D 86.7%, and phallus C 28.9% [( F (12)=61.66, P <.0001]. The effects of semen viscosity and phallus type were examined using a 3 2 factorial design. As shown in Fig. 6, there was a main effect of Phallus Type [ F (2,29)=108.15, P <.0001], a main effect of Semen Viscosity [ F (1,29)=13.49, P <.001], but no Phallus Type Semen Viscosity interaction.