All bees were trained using both antennae and tested for recall using either the left (LA) or right antenna (RA), the other being covered with latex. The mean number (±standard error) of type A responses (proboscis extension response, PER, to lemon odour and not vanilla odour), B (PER to both odours), C(PER to vanilla and not lemon) and D (no PER) are plotted. The A scores were analysed by GLM using the factors Antenna (left versus right antenna in use) and Order (left antenna tested first or right antenna tested first). At 1–2 hours after training, there was no significant main effect of Order (F(1,20) = 2.896, p = 0.104) and no significant interaction between Order and Antenna (F(1,20) = 0.478, p = 0.479). The main effect of Antenna was significant (F(1,20) = 7.358, p = 0.013). A responses were significantly higher when the right antenna was in use than when the left antenna was in use. At 23–24 hours after training, there was no significant main effect of Order (F(1,22) = 0.124, p = 0.728) and no significant interaction between Order and Antenna (F(1,22) = 1.597, p = 0.220). The main effect of Antenna was significant (F(1,22) = 14.064, p = 0.001). A responses were significantly higher when the left antenna was in use than when the right was in use.