I can not understand the meaning of following discription, explain it with graphs(I can not imagine a graph.)

This assumption may be checked visually by plottingy= residuals versusx= fitted (estimated expected) values. Constant variance will show up if the graph forms a fairly level band across the page. A scatterplot of residuals versus fitted values can be produced inRusing theplotfunction applied to the fitted values (as the first argument to the function) and residuals (as the second argument) for anaovobject.

If, for example, the variability of the group's data is larger for groups with larger means (as happens with lots of biological data), this will appear on the graph as a widening of the band towards the right. In this case a log transformation of the data is indicated. (We will use the natural log, to basee.) The observed valuesyare all replaced by log(y), the ANOVA is done again, and the diagnostic graphs for checking normality and constant variance are again plotted. A comparison of these diagnostic graphs with those for the raw data will help indicate if log data should be used in the final analysis.