This synthetic data-set was constructed by the
statistican Frank Anscombe to illustrate how
residuals can be used to discriminate between
data-sets that give identical fitted regression
models.

The six columns correspond to:
   1.  x=explanatory variable for each of responses 1, 2 or 3
   2. y1=response 1
   3. y2=response 2
   4. y3=response 3
   5. x4=explanatory variable for response 4
   6. y4=response 4