Outliers or clusters must be very distinct before we should conclude that they are real, in the absence of further external information confirming that the individuals have different characteristics. This is particularly true with small data sets where outliers, clusters and other patterns may arise simply by chance.
Be careful not to overinterpret features in scatterplot unless they are well defined, especially if the sample size is small.
Sweet corn size