This is slightly off-topic, but I found a couple of articles dealing with the inheritance of orange (pelargonidin derivatives) in crosses with yellow varieties.
de Vries: Flavonoids in rose petals (1974)
http://bulbnrose.x10.mx/Roses/breeding/ ... e1974.html
de Vries: Rose pigments II (1980)
http://bulbnrose.x10.mx/Roses/breeding/ ... s1980.html
One of the disadvantages of "old" research is that some of the useful facts have not yet been completely separated from the not-so-useful bits. Still, there are some worthwhile observations.
First, there is a correlation among the occurrence of kaempferol, callistephin
(pelargonidin-3-monoglucoside) and pelargonin (pelargonidin-3,5-diglucoside).
Likewise, quercetin is associated with chrysanthemin
(cyanidin-3-monoglucoside) and cyanin
This may look very esoteric, but it has practical applications. A white or yellow rose with no trace of red pigment, but containing quercetin, will likely give red or purply offspring when crossed with an orange type. A different white or yellow, containing kaempferol, will give more orange-toned offspring when crossed with an orange.
The 1974 paper gives a long list of roses with their flavonoid contents. Some of these, identified as "IVT no ..." were bred by the authors but not introduced. They did contribute to the statistical analysis, so there's that. Some of the others might still be useful for breeding.
Arisumi (1963) also dealt with the various anthocyanin pigments, and raised an interesting issue. Even in the dwarf Polyantha roses that can produce lots of pelargonin, callistephin is rare.
http://bulbnrose.x10.mx/KKing/RosePigme ... 963_4.html
In most other plants, the 3,5-diglucosides are produced from the 3-monoglucosides. Not in roses. Roses are weird. Instead, there is a separate enzyme that sticks a glucose molecule on position 5 of the anthocyanidin molecule, which is an unstable thing. The same enzyme then quickly sticks another glucose on position 3, and all is fine and stable.
The big point here is that the 3-monoglucosides and 3,5-diglucosides are chemically akin, but result from different enzymes (encoded by different genes).
This all becomes a lot more interesting and a little more complicated because the 3-glucotransferase (3-GT for short) comes in at least two flavors. One works in the usual way, making pigment in the petals that are available when the flowers open. The other form just sits there waiting for light and heat before it gets busy. This is the enzyme responsible for the lovely chameleon roses that deepen in color as they age. 'Mutabilis' is an example, 'Color Magic', 'Emmie Gray', and 'Masquerade' are a few more.
Sadly, this same 3-GT is responsible for the faint purplish flush that dulls the color of 'Margo Koster' in the summer months.
Stable coloring is good. But there is also the possibility of deepening color of one pigment compensating for the fading of another.
For instance, luteolin deepens in color (yellowish) when exposed to UV. Yellow carotenes tend to fade. I don't know for sure that 'Devoniensis' contains luteolis, but the flowers do become more yellowish (yellow snow-ish) as they age. This color is not very attractive, but it wouldn't be bad if it were combined with bright yellow.
http://bulbnrose.x10.mx/Roses/Rose_Pict ... ensis.html