The Rose Annual pp. 11-17 (1956)

Roses at Bayfordbury
(The National Rose Species Collection)
Gordon Rowley

*The Rose Garden, Edn. VIII, 1881, 29.

In 1846 we read of the setting up of a formal rose garden surrounding a pool and fountain between the stables and kitchen garden. We know little of the original roses grown there, but the collection achieved an honourable mention by William Paul*—no mean compliment in a county renowned for its famous rose gardens. When the Agricultural Research Council decided in 1946 to establish Species Collections as permanent reserves of plant-breeding material, the new home of the John Innes Institution was chosen for the genus Rosa, and by April 1951 the Rose Species Collection was complete in broad outline (Fig. 1). But no plant collection is ever complete as long as new seedlings are germinating in the wild, and the search for new material and the discarding of old continues.


The first problem to attract attention, apart from the unending worries of names and identities, concerned the part these species had played in the development of the garden roses of today. A search was begun to discover as many as possible of the "missing links" between, for example, the wild Musk Rose of the Himalayas (R. moschata) and its modern descendants like 'Wilhelm' and 'Will Scarlet'. Prying into the past is a thrilling and never-ending sport: it brings disappointments and pleasant surprises. How exciting it was to be able to plant side by side 'La France', the first authentic hybrid tea, 'Cheshunt Hybrid', the first English Hybrid Tea (retrieved by luck from a garden in Gloucestershire) and 'Soleil d'Or', the forerunner of the Pernetianas (Fig. 7). Picture the delight at receiving a plant of 'Madame de Tartas', that famous old Tea Rose of 1859, parent of 'Caroline Testout', and source of so many favourite early Hybrid Teas! This was smuggled through the Iron Curtain by a well-known rosarian who spotted perhaps the only surviving source in the rose collection at Sangerhausen. Here these forgotten beauties can flourish without fear of the extinction which so readily overtakes once-popular roses. The catalogue of resuscitations could be extended, as well as the list of roses still sought in vain: 'Lady Mary Fitzwilliam', 'Lyon Rose', 'Hume's Blush China', 'Park's Yellow China', and so on. The greatest fillip to this search has come from the firm of Messrs. T. Hilling & Co., who presented us with nearly 400 of Mr. G. S. Thomas's collection of Albas, Damasks, Mosses, Chinas, Bourbons, and other shrub roses. Where better to plant these than in the original Bayfordbury rose garden of 1846? (Figs. 5, 6).

The living museum grew and grew, and in 1951 a separate planting was made to show the stages in the development of the modern Floribundas through first the Chinas, Bourbons and Noisettes, then the Portlands, Hybrid Perpetuals and Teas, and later the Hybrid Teas enriched by the Pernetianas and Dwarf Polyanthas (Figs. 2, 7). The side branches of this family tree are devoted to the origin of minor groups like the Scotch Roses, Hybrid Musks, Hybrid Rugosa, and so on.

The first outcome of the study of rose pedigrees has been presented by Miss A. P. Wylie,1,2 who based her phylogeny on the cytology of modern roses and their predecessors. A continuation of this work, involving numerous hybridizations between the ancestral species in the attempt to retrace the early steps, will take some years before conclusions can be made. It is part of the general rose-breeding programme which aims at finding out how the species can contribute directly or indirectly (i.e. as rootstocks) to the betterment of garden roses. In this connection it is interesting to note how few of the species (as little as 5 per cent) have participated in making the big groups of garden roses. All can be traced back to about seven or eight wild species—often to single clones of species without regard even for their natural variation. Put differently, our shallow genetic pool is long overdue for a high tide to renew its stagnating contents. Many breeders, sensing this, are already busy tapping new sources like R. macrantha and R. multibracteata, but understanding the nature of a problem goes far toward providing a solution. The rose species offer many features that have been overlooked in the scramble for bigger and brighter blooms: good foliage (R. bracteata, rubrifolia, fedtschenkoana), shrubby habit (R. nutkana, moyesii), ornate prickles (R. sericea pteracantha) or, if you prefer, lack of prickles (R. blanda), disease-resistance, hardiness, long life, and so on. And for the seeker after yet further colour novelties, there is the unsolved riddle of the untameable Hulthemosas.3

As a single example of the introduction of "new blood" let us look at a planting commencing with the wild English sweet-briar, R. rubiginosa. By a crossing using the pollen of a garden rose, probably a Gallica or Hybrid Perpetual, Lord Penzance produced 'Lucy Ashton' in 1895. Twenty-one years later a seedling from 'Lucy Ashton' gave 'Magnifica', a free-blooming, nearly double, vigorous shrub still retaining characters of the species in its scented foliage. 'Magnifica' has been the means of handing on new characters of hardiness and disease resistance into the modern Hybrid Sweetbriars, beginning with 'Rosenwunder' in 1934, and brought fully up to date in roses like 'Florence Mary Morse', with the Floribunda-type habit, vermilion colour (from 'Baby Chateau') and long flowering period. A more recent line, from Kordes, and promising much for the future is a cross of R. kordesii and 'Eos', itself a hybrid between R. moyesii and 'Magnifica'. Thus elements of at least five rose species are combined in a single genotype: R. gallica, a tetraploid, and R. rubiginosa, a pentaploid from Europe: R. moyesii, a hexaploid from Western China, and R. rugosa and R. wichuraiana, diploids from the Japan-Korea area.


The botanist can learn much from studying collections of wild species grown side by side, especially if self-pollinated seedlings are there for comparison. It sets the table for a taxonomic feast far better than the canned fare of the herbarium. Where Rosa is concerned it makes one wonder at the achievements of past taxonomists like Crepin and Boulenger who worked on what was by analogy a starvation diet. The study of fragmentary specimens led to a plethora of names, and to the invention of numerous "hybrids" to bridge gaps between allied species. Many of these names can no longer be typified or were described in too little or too great detail to merit retention. The alleged hybrids, such as R. x macrantha, x malyi and x spinulifolia, can now be put to the test by a study of their meiosis and pollen fertility, as well as by trying to remake them from their supposed parents. The wild roses of Great Britain are especially in need of this type of study, as most of them belong to the Caninae or dog roses with a unique subsexual type of breeding mechanism. Reciprocal hybrids between the same two species look quite unlike each other, and combine different proportions of parental characters. Aneuploids—that is, deviants from the normal regular multiples of the seven chromosomes—frequently occur.

When a nurseryman advertises "Six choice bedding roses for 24s." this is, strictly, not what he means at all. What he should say is "Twelve half-roses for 24s.", because each "rose" is compounded of two parts: a flowering above-ground portion which is privileged to bear the name wedded to a root system of an entirely different rose—usually one of the wild species or something not far removed from it. To this symbiotic relationship of a successful stem on a successful root we owe much of the extraordinary popularity of roses today. Without it many choice blooms would have languished in the seed-bed, and distribution of others would be intolerably slow and expensive. I stress this duality of garden roses because it is easily overlooked, and leads to anomalies like selections of garden roses for growing on chalk or light soils or in dry places: about as sensible as offering a hen lump sugar to make it lay square eggs.

If we probe deeper to find out which rootstocks are best suited to chalk or light soils, or dry, we find a very different story. Little or no work has been done on which the experimenter can rely—certainly nothing compared with the extended survey of fruit stocks undertaken at East Mailing. Assessing the merits of different stocks is not a matter of collecting growers' testimonials (of which there are already far too many) but of extensive, long-term field trials in which stocks and scions are compared side by side. But whereabouts can one begin, and which roses should be selected for trial? The forty varieties of stocks at Bayfordbury take no account of the many local races favoured by nurserymen and worked up from supplies in their own area. Multiply this figure by the many thousands of garden roses, all potential customers for borrowed roots, and the number of stock-scion combinations is enormous. In turning to the search for new stocks, new problems arise. First, both seedling and clonally propagated stocks are in common use and each is claimed to have its own merits. The breeding programme differs for both. A seedling stock is required to germinate freely and to breed true for all characters influencing vigour, length of life and compatibility. A stock propagated vegetatively can make use of hybrid vigour, but must root readily from cuttings. Being genetically uniform throughout, clonal stocks are all equally susceptible to attack by fungus or virus diseases. Again, the most promising new stock may prove to be incompatible when budded with some garden roses, and this can be found out only by expensive and time-consuming field trials. Clearly we must not hope for quick results or one panacea to end all budding problems.

The first rose stock trials at Bayfordbury are on a small scale and confined to a simple comparison of widely grown roses. They were planted in the autumn of 1954. The firm of Samuel McGredy & Son presented us with 384 maiden plants consisting of eight hybrid teas budded on eight stocks in all possible combinations. These have been planted on a medium loam soil in four randomized blocks with a guard row round the edges (Fig. 4). Comparisons are made by recording the vegetative performance of each as annual height, number of shoots and weight of prunings, and the floral performance as number and quality of blooms. The last-mentioned character is much harder for the scientist to assay than the poet and painter, whose rich vocabulary unfortunately fails to satisfy the statistician in search of mathematically comparable data. Hardiness, length of life, suckering, incidence of Mildew, Virus, Black Spot and other rose fanciers are also noted for comparison later. What can we hope for from such a trial as this? At most a recommendation in five or six years' time of the best stock for each of the given eight roses if grown under similar conditions, the stock with the best all-round performance, and perhaps a few observations on the relation between plant size and freedom and quality of blooms. The time and length of flowering season may also show differences of importance to growers. Later trials will compare performances on two contrasted soils: a light gravelly loam and a sticky calcareous clay. This is aimed at putting to the test the popular choice of R. canina for heavy soils and R. multiflora or rugosa for light.

Propagation of rose stocks is likewise receiving attention. A stratification technique has been developed for saving a year's wait in R. canina4 and is worth trying for other refractory cases. The failure to root uniformly from cuttings can be tackled in various ways. For instance, trials were made of...

<missing text>

... light waves. Some may be so, but I can imagine few workers on Rosa devoid of a certain pride in their lucky choice of subject, whether the brilliant, shapely double blooms representing perhaps the summit of man's achievement in flower-breeding, or the no less perfect wild, five-petalled rose as Nature bred it. There is indescribable joy in watching one's own hybrid seedlings bloom for the first time—joy that compensates for hours of failure and fruitless effort. Each seedling defiantly disproves that "there's nothing new under the sun". And the pinning down of a single fact more than makes up for the time taken counting and calculating, when life's "bed of roses" seems to contain nothing but the prickles.


  1. WYLIE, A. P., "The history of Garden Roses part 1", in J. Roy. Hort. Soc., LXXIX, 1954, 555- 71; "The history of Garden Roses part 2" LXXX, 1955, 8-24, 77-87.
  2. WYLIE, A. P., "Chromosomes of Garden Roses", in Amer. Rose Annual, XXXIX, 1954, 36-66.
  3. ROWLEY, G. D., "Hulthemosas—New Hope for the Rose Breeder", in The Rose Annual, 1955, 37-40.
  4. ROWLEY, G. D., "Germination in Rosa", in 44th Ann. Rep. John Innes Hort. Inst., 1953, 27-8.
  5. ROWLEY, G. D., in 45th Ann. Rep. John Innes Hort. Inst., 1954, 10-11.
  6. GARNER, R. J., in "Propagation by Cuttings and Layers", Imp. Bur. Hort. & Plant Crops, Techn. Bull. 14, 1944, 80.