In your experiment, random selection would occur at the point where you determine which seeds to plant in the pots of soil and sand. If you were to randomly select which seeds to plant in which medium, for instance by using a random method (like drawing lots or using a random number generator), that would constitute random selection. This method would help ensure that any variability in growth is due to the conditions of the soil versus sand, rather than any bias in the selection of seeds.
Additionally, if you were to conduct trials with multiple sets of seeds (for example, using more than 10 seeds), random selection could also apply when choosing which seeds to use for each trial to ensure a fair comparison between different growing conditions.