In some cases, ErasmusLadder trials which are planned as containing perturbations may not actually include them. In order to present a perturbation at the correct time, ErasmusLadder predicts when the animal will reach each obstacle based on its position and speed. In some cases, especially if the animal's progress is unsteady or inconsistent, ErasmusLadder may not be able to make that prediction and the perturbation will not occur during that trial.
The default Erasmus University protocol uses a total of 42 trials in order to get enough trial data for all three protocols (US Only, CS Only, and Paired) with a prediction. The initial development at Erasmus University emphasized this so that there is enough statistical power, even if only a fraction of the trials has a prediction. This is, for instance, the case in very ataxic mice. Originally, the protocol contained 72 trials, but this was later lowered to 42 to reduce fatigue, while maintaining sufficient statistical power.
If you do not design your protocol with enough of each trial type for statistical analysis, you will likely need to increase the number of trials. Keep in mind, the exact number of each trial type is not the goal of the overall goal of a protocol, it is designed to get just enough of each trial type to be able to calculate the statistics.