We list Mendel's data. Each part can be examined for goodness-of-fit to the theoretical proportions. They can be aggregated into chi-squares, or left as single d.f. terms.
i) 3:1 ratios for 2 seed and 5 plant characters.
ii) Plants 1.....10 in experiments 1 and 2.
iii) As for i) but with the 2:1 ratios.
iv) As for iii) assuming 10 plans tested each time.
v) Dihybrid cross data.
vi) Trihybrid cross data,.
vii) Backcross data.
If you are lazy, look at Edwards' paper where everything is organized into 1 d.f. contrasts, and the set of all such combined into a normal qq-plot. If you take this route, at least do: the following:
Exercise 1. Prove that the 1 d.f. terms in the decompositions of the frequency tables are asymptotically independent. E.g. if w, x , y and z should be proportions 9:3:3:1 under the null hypothesis, then w - 3x - 3y + 9z, w + x -3y -3z and w -3x + y - 3z are asymptotically independent.
Exercise 2. Choose a plant at random from those exhibiting the dominant form of a trait in the cross of first generation from the hybrids, and examine 10 of its selfed progeny. What is the chance that it will appear pure breeding?