Bayers therom by quantitative analysis.

Has anyone ever tried to use statistical analysis to create ideas for crosses? It seems enough data today exists. Meteoroligists and Oil companies use it why not us?

The only thing that I’ve seen so far with any statistical data on roses is in the below link. Through all my wandering over the internet I have yet to see anything on disease resistance, petal count, etc.



I used the MR XI database to do some statistics a few years ago and had a couple of articles published about it in The American Rose. One experiment that I tried but didn’t write about in the articles was to get ideas for crosses for fragrant roses. I did statistics on the fragrance of the first-generation descendants of all of the roses in my garden. I found that the rose that produced the highest percentage of intensely fragrant seedlings when used as a seed parent was Fragrant Cloud, and the rose that produced the highest percentage of intensely fragrant seedlings when used as a pollen parent was Lavender Pinocchio. So I did the cross Fragrant Cloud X Lavender Pinocchio. That is a cross that would have never occurred to me if I hadn’t done the statistics. I wouldn’t have expected the seedlings to be very healthy, and they weren’t. On the other hand, many of the seedlings were very fragrant. I kept one seedling from the cross and it is the most fragrant rose I have. It has beautiful red blooms that nod like tea roses. It is susceptible to mildew, but I kept it anyway because it proves that the statistics worked.

One of the reasons I originally used ‘Sweet Afton’ was because of the large percentage of first generation fragrant offspring. It also happens to sets hips with practically any pollen and the seed germinate so I ended up with many offspring. Health of descendants could be better but at least I generally have something fragrant to work with.

‘Fragrant Cloud’ never produces hips here but I recently produced a descendant through ‘Just Joey’ and it is very fragrant.

There is a slight propensity to Powdery Mildew.


Hi Wade, I used to use Modern Roses and the ARS information on new roses - looking at parentages, to try to select “good parents”. I literally spent hours pouring over the information. Though the information there is somewhat useful, it can be misleading. I have seen others use those sources to decide whether a rose would be “good” or “bad” to use in a cross, or to try to determine propensity for bloom color with certain crosses, using statistics derived from those sources. The problem is, that you are only seeing the surviving seedlings that were ultimately registered. That information tells you nothing of the number of seedlings discarded in the particular crosses, or about their bloom colors (or about any of the other numerous traits of interest). For that information to have the most usefulness, you would need to include information on all of the seedlings of a particular cross. That is certainly possible when doing statistical analysis on your own seedlings, from your own crosses - that information can be most useful.

Though these sources (MR11, etc.) do have a wealth of information and may be suggestive for directions to go in your breeding program, I have not found them to have any real statistical power. I wouldn’t discourage anyone from using those resources, however, I would be cautious recommending them as valuable for real statistical analysis - too much important data is missing.

Using roses that I like that have good fertility, and important qualities that I want in my seedlings has been most helpful to me.

Jim Sproul

Jim, the point you make about only seeing the surviving seedlings that were ultimately registered is one of the reasons I was asking for “whole litter” information for the color project I want to do. It isn’t so much the colors one gets from a cross that counts, it’s the proportions of colors/spots/disease resistance/whatever that matters.

I wondered myself at times whether a person could predict the flower size of any given size + or - a certain percentage. I wondered this when I was more active in breeding Daylilies. I know they can predict in many orchid crosses the resulting size of the seedlings without it ever flowering before. I was always curious but I thought that the tetraploids and dwarf daylilies would throw of the data. I was reminded of this train of thought reading this thread. I would imagine that with roses the same two problems would arise. Perhaps one day when I have the room maybe I will start with some diploids and see if their is a formula somewhere that works out. This I would do more out of curiosity. If I where a cat I would be dead by now!