IF YOU HAVE a rare plant, not necessarily first-rate but one you would not know how to replace, it is always a good idea to progagate it and establish it in two or three places about the garden. It is also an excellent idea to give starts of it to friends.

I recently gave a magnificent clump (if I do say so; after all, it just grew and I can take no credit for that) of the wild white Iris tectorum, the roof iris of Japan, to a friend in New England.

He seemed no end pleased, as well he should have been, and thought I was wonderfully generous since I gave him half of my small stock. But I was not generous, merely prudent.

If my iris should die or for some reason vanish, it is possible his will still be flourishing and he might give me a new start of it.

This iris is quite remarkable, and I feel guilty that I have done nothing to see it distributed in general commerce, but its story may interest you.

About 45 years ago my friend Eddie Fox of Memphis decided that a pink variation of the wild lavender I. tectorum would be nice to have.

He knew there was a wild variation of this iris that is pure white with canary-gold touches at the crest. He also knew that the wild lavender had been crossed with a garden iris (a diploid tall bearded, in case any breeders are with us today) and the resulting seedling was called 'Paltec,' alluding to a cross between I. tectorum and I. pallida.

So he thought, reasonably enough, that if he crossed the wild white with a diploid pink tall bearded he might get a flower like 'Paltec' only pink.

He chose various pink irises for his project, especially the one called 'Dog Rose,' and year after year he put the pollen of the pink garden iris on the sticky style-arm lip of the wild white, hoping against hope to get a seed pod.

The cross is a difficult one, since the pink garden iris and the wild white roof iris are from utterly different sections of the iris genus. The white one does not even have a beard (as all our ordinary tall bearded irises do) but instead has a crest, like a small golden cockscomb.

But since at least one successful cross had been raised, namely, the lavender 'Paltec,' he knew such a cross was possible.

Imagine his delight when his first attempt produced not merely a seed or two but dozens of seed pods; all with fertile seeds. He planted these outdoors in October and a couple of springs later a whole bed of them bloomed.

Every single one of them white, exactly like the female parent, I. tectorum. Undismayed, he crossed these seedlings among themselves, and also crossed again with the pink male parent and again waited for the seedlings to bloom.

Again, they were all white, indistinguishable from the wild I. tectorum.

Generation after generation he did this, each year making more crosses with pinks and sowing the seeds. He must have done this for 30 years. To the very last he got nothing but pure whites, exactly like the wild roof iris.

Someone finally suggested he make the cross the other way; putting the pollen of the wild white on the female organ of the garden pink. He did so and got no seed pods. He persisted and at last got one seed pod. Not a very large pod, but one with few seeds that looked a bit odd. Of these, I believe four or five grew to maturity and flowered.

There was not the slightest trace of I. tectorum in them. They were perhaps the ugliest irises I have ever seen. They were on the short side (which I thought was encouraging, since I. tectorum is only a foot high) and the colors were ghastly, so was the shape of the flowers; and yet there was no clue that they bore any genes at all from the father plant.

I thought they should have been intercrossed and another generation or two raised from them. For while I. tectorum did not seem evident in the first generation, the flowers were so ugly I could not believe they came from the pretty, elegant 'Dog Rose' entirely. I felt that these ugly flowers had inherited something from the wild white; though it did not show in the first generation.

Unfortunately, the ugly seedlings were somehow lost, and that line was not continued. My friend, Eddie, sad at the very ugly results of using the pink iris as female parent, returned to his original practice of using the wild as the female parent.

I cannot say how many generations he raised, maybe 25 or 30, each year using the wild white as female parent.

It was clear to me that while the wild white was setting fertile seeds like mad, no actual cross had occurred. No inheritance was being passed from the pink parent.

This is not too surprising, since I. tectorum is highly fertile to its own pollen. If you never put pollen on it at all, it still sets seed abundantly in some years.

Whether this is because it fertilizes itself before the flower opens (which I do not think is true) or whether bees get to it, fertilizing it with its own pollen before the human breeder gets to it (which is likely) I do not know. It is also possible, so far as I know; that the pollen from another iris stimulates I. tectorum to set fertile seed, without any inheritance whatever from that foreign pollen.

There are plants, fishes, and perhaps other creatures that do not have offspring unless foreign pollen or sperm is applied, but which nevertheless produce offspring identical with the female parent. I thought this was happening with Eddie's white tectorums, otherwise in all those years there should have been at least some detectable variation from all those crosses with tall pinks. But to the very last, the progeny were identical, not differing in even the slightest detail, from the wild white he began with decades before.

So the project was a failure, in the sense that Eddie never raised his pink tectorum.

In my opinion, at least, it was a great success in another way:

The wild white, which everyone knows is one of the most delicately beautiful of all flowers on earth, has always been a bit miffy in gardens. It likes plenty of humus, plenty of moisture and plenty of good drainage.

I grew it once (just to see what happened) on a rotted straw archery butt, and it flourished admirably. Just as it is said to have done in the past on the straw roofs of Japanese huts.

But it soon exhausts the soil in some way. If the gardener does not keep an eye on it, and move it to new ground every two or three years, it often peters out and dies away. Also, in gardens the wild white simply is not as robust as the wild lavender tectorum, and nothing like the robust and persistent early diploid tall bearded irises or the old triploid intermediate irises like the deep purple-blue one you see everywhere in the spring.

What Eddie Fox was in fact accomplishing was a new strain of the wild white tectorum. As generation followed generation, the strongest of the wild whites lived and the others either died or came into flower a year late or were not robust-looking plants. So one way or another, the most vigorous ones were bred from, generation after generation.

Fortunately, the flower in all those generations has not varied a jot or a tittle. I say fortunately, since it is hard to imagine any change in any direction that would not decrease, rather than increase, its beauty.

At the last, Eddie's iris friends began to notice that Eddie's white tectorums behave a lot better than the usual white tectorums. If they were briefly overrun by Bermuda grass in the summer, or if they were accidently mowed with a grass cutter, they often survived, whereas a gardener would not expect the somewhat fragile white tectorum to recover from any major setback.

I do not think my old friend, now with God, was ever making the slightest advances towards his goal of a pink tectorum. But what he was in fact doing, without even intending it, was raising generation after generation of whites, selecting (unconsciously) in each generation the sturdiest garden performers, so that after many years he possessed a strain of white tectorums better than any in commerce.

I am not sure of that. I have not taken white tectorums from commercial sources and grown them along with Eddie's whites and kept records of their performance over the years.

But I and other iris nuts were convinced that Eddie's whites did not die out, did not rot, did not freeze to the ground so often in the spring, and did not require any such pampering as the white tectorums we used to grow before.

And it stands to reason we did not simply imagine this, though none of us ever actually tested it. Plants that come true from seed (like the tectorum irises) are much better raised from seed over the years than propagated from offsets.

If you propagate asexually, as from offsets, you perpetuate any virus or any other weakness that individual iris may have. If you raise from seed, then select parents, then raise from seed, and so on for a few dozen generations, it would be surprising if you did not wind up with a superior strain as far as garden performance is concerned, since you would have automatically selected out the poor performers.

Over the years, in moving, I somehow lost my stock of Eddie's White, as I call it. But fortunately, I had given some to friends.

They got me started again. This is the basis (for I am shamefully greedy and stingy by nature) for my generosity in giving the great clump to my Connecticut friend.

Someday I hope I shall get collected sufficiently to propagate Eddie's White more than I have done, and may build up a good stock and persuade some iris nursery to raise it and sell it to gardeners generally.

It seems to me a very fine thing, and never mind the pink (I think and always did think a pink 'Paltec' would be about as ugly as the lavender 'Paltec'), to take a somewhat delicate wild white iris and convert it into an ironclad garden plant, without altering in any way the fairly supreme beauty of the original wild blossom. Eddie meant to turn it pink. God in his grace denied him that (perhaps foolish) goal, but gave him a better one that he achieved without even being aware of it.