Taking multiple readings allows for errors (known as anomalous data points) to be spotted.For example if you were to take just one reading in an experiment and this was wrong, there would be no way of telling this and the you would record the wrong answer as right, whereas, if multiple readings are taken, it is easy to spot if one of them doesn't fit the trends shown by the others, and this can be discarded without affecting the results. Additionally, if multiple readings are taken and these all fit the same trend, this shows that the results are reliable.