I once posted a message quoting what Ralph Moore once told me, saying that he believed that in a sample of 100 seedlings, you should get to see the full scope of what a cross will do. Of course, the number 100 is fairly arbitrary; if you have 90 seedlings you will still get a very good perspective of the value of a cross, and 200 seedlings will surely give you an exhaustive overview. Personally, I like to aim for 150 seedlings per cross.
The next question then becomes, how many pollinations were required to arrive at that ideal sample of 100 or more? Well, that depends a great deal on the plants you are working with. If you know that Seed Parent X produces full hips with many seeds, most of which germinate, then you can make relatively few pollinations and still arrive at your ideal sample. For me, roses like ‘Sequoia Ruby’ will give me a crop of 100 seedlings or more with as few as 20 pollinations. ‘Little Darling’ comes close to that rate as well, as does ‘Sequoia Gold’. Parents like ‘Abraham Darby’ are not so generous, requiring more like 40-50 pollinations to achieve a sample of 100 seedlings. Some parents that I work with are very hesitant to produce viable, easy-to-germinate seed and so, I have to make 1 pollination for every seedling I hope to get. Naturally, I avoid working with such reluctant parents unless they offer some very special opportunities for advancement of my work. ‘Muriel’ is one such rose. Its pollen will set seed only rarely and only on certain plants, and seed usually isn’t all that readily germinated. My guess is that the rate of success is about one seedling or less per pollination.
As for evaluating results, I believe that if you have a large sample of a cross (100 or more) and you do not see anything worthwhile in that sample, then you are not likely to see anything better by increasing the sample size, or repeating the cross the next year. If I get nothing I like from a cross in 100 seedlings, I won’t make the cross again, absolutely. If I did like what I saw, and if the scope of variety in the cross was very broad, I will undoubtedly make the cross again to have more opportunity to select something good. Two years ago I made a cross of ‘June Laver’ X ‘Out of Yesteryear’, and it resulted in over 70 seedlings, of which I kept about 20 for evaluation. Thats a very high percentage of keepers! In 2004, I narrowed that selection down to three seedlings, and next year I suspect one more will be eliminated. This year, I remade that cross to have a chance at finding something else I really like. The cross gave a very broad spectrum of variations, and so another sample would possibly reveal more than I saw the first time. In such a case, it is well worth repeating a good cross.
The bottom line: it helps to know how fertile your seed parents are. For me, it helps to do a preliminary test with a prospective parent before I attepmt any serious work with it. I make a few crosses, with anything I have that is a known fertile pollen donor, and I germinate them and evaluate. There are three things that I can determine in this initial assessment: 1) does the plant form seeds readily? 2) do the seeds germinate easily? and 3) is there anything of value in the test sample offspring? These three things can be determined in a test sample from an un-tested seed parent. If I have positive answers to all three questions, I will likely work with that rose in earnest the following year.
I hope that provides some kind of answer to your question.