J Hered (1938) 29(4): 115-121.
Forman T. McLean

William Herbert’s "Gladiolus fragrans"
Vase of first-generation hybrids between Gladiolus tristis and G. recurvus, two fragrant species related to garden gladiolus. They are very different in respect to odor,—tristis having a lily-like scent at night, while recurvus has a mild violet-like scent during the day. All of the first generation hybrids have the violet fragrance of G. recurvus, but lack the tristis fragrance, though it reappears in some of the seedlings in the second and later generations. This cross was made over a hundred years ago by Dean William Herbert, one of the great hybridists of the early nineteenth century, who called the hybrid Fragrans. The hybrid was propagated for a time, but eventually disappeared.


Figure 1
Gladiolus tristis concolor (A) has a fragrance of night jasmine, which is recessive in hybrids with G. recurvus (B). The violet day-scent of the latter species is dominant in inheritance and is transmitted to all the hybrids. These species are diploids, while the garden gladiolus is a tetraploid, and produces only sterile non-fragrant hybrids with tristis and did not cross with recurvus. However, in the cross between the garden varieties and the tristis x recurvus hybrids, some seedlings are found which are fragrant and have flowers of fair size and of striking color.

When this project, to put fragrance into the race of garden gladiolus, was started fifteen years ago6,7,8, about a half dozen fragrant wild species of the genus were known from South Africa; all slender, delicate, winter-growing forms, more resembling the greenhouse freesia than the gladiolus as we know it. The dream of a stately gladiolus with fragrance added was not new. One of the wild fragrant species, Gladiolus tristis, is grown as a curiosity and to some extent commercially in England and Holland; and gladiolus breeders have tried sporadically for an hundred years to get its powerful but elusive night fragrance into more acceptable garden gladioli. The crossing of tristis with garden varieties proved easy, but the results were highly discouraging. The hybrids were small, sterile mules,—and scentless.

So, the problem is a two-fold one: to find or produce a fertile fragrant race of gladioli, that will transmit fragrance to hybrids between it and scentless garden varieties; and next to develop acceptable garden hybrids with fragrance.

At the start, three fragrant species were available: (1) Gladiolus tristis, the little, slender greenish-cream colored, night-scented Avendbloem of the Boers; (2) the closely related related G. grandis, the Brown Afrikander; and (3) G. recurvus, the Mauve Afrikander, which has a mild, violet-like day fragrance and narrow little flowers

Of these, G. tristis had already been crossed with garden varieties without success; G. grandis proved to be similar, but inferior as a parent; and C. recurvus failed to cross with the scentless garden sorts. Thus the direct attack on the problem failed as the crossing of fragrant species with garden varieties did not eventuate in any fragrant hybrids.

There still remained the possibility of building up a fragrant hybrid race from fragrant species, hoping that the hybrids would, like G. tristis, cross with the garden sorts, and unlike that species, might transmit fragrance to the resulting progeny. The best prospect seemed to be a cross between G. tristis and G. recurvus. This same cross was made by Dean Wm. Herbert, of Manchester, England,4 over an hundred years ago, and he called his hybrid Gladiolus fragrans, which was encouraging for the present purpose: but he apparently did nothing further with them, being more concerned with other brighter colored hybrids. None of his G. fragrans appear to have persisted beyond the middle of the century.

In the present study this cross was repeated and the hybrids interbred, to learn the manner of inheritance of fragrance and of such other species traits as might prove to be a benefit or a hinderance in hybridizing with the garden sorts for fragrance. While the purpose was candidly a practical one, — to produce acceptable fragrant hybrid gladioli, — because of the hundred years of failure in similar efforts by others, a detailed study was deemed necessary. The results appear to justify this procedure, as well from a practical as from a scientific standpoint.

Crosses were made using as seed parent one clon of the wild species here designated Tristis II, pollinated by two clons of the other wild species, called respectively Recurvus Bronze and Recurvus Violet Purple. The records of the progeny of the two groups were kept distinct. This was necessary, since both species studied are quite variable in color and markings, though constant in their diagnostic characters, which include flower fragrance. Then the F1 hybrids from Recurvus Bronze were intercrossed, and also backcrossed with Tristis II. In all 2,200 hybrids were studied in some detail, recording not only their fragrance, but 23 other characteristics, to learn if these might be linked to fragrance and prove beneficial or detrimental in the subsequent hybrids which it was expected to use with the garden varieties.

Noses Different

The study of the inheritance of fragrance was beset with problems from the start. First, there were no clear-cut standards for measuring fragrance. We must depend upon our noses. Then, individual noses record widely different results, as Dr. A. F. Blakeslee has recently emphasized.2 All of the observations of fragrance and of other traits used in this study were made by the writer, with the help of two coworkers. Miss Lunt and Mrs. Jennison, whose perceptions of fragrance tallied very well with his. A notable fact, that stood out whenever other people were called in to test the flower fragrance, was that tobacco smokers quite regularly had less sensitive noses than non-smokers. The writer is not a smoker, which may be one good reason for his undertaking this particular study. Even so, the results are presented only as the perceptions of one nose, and this will account for the rather consistent findings. It was evident throughout the study that any attempt to have averaged the observations of even a large number of people would have been unsatisfactory, individual variations in perception of fragrance being too great.

To show the mechanism of inheritance of fragrance in these hybrids, the observations of fragrance in the two groups of F1 hybrids, in the F2, and in the backcross to Tristis II are summarized in Table I. The fragrance is tabulated under eight items. First comes Non-fragrant, of which there were a surprising number in the F2 and the backcross, considering both species are consistently fragrant. The simplest explanation of the Non-fragrant (Line l) F2 and backcross individuals is to assume that there are non-fragrant allelomorphic factors to those for violet-like Recurvus day fragrance (Line 2), and for the lily-like Tristis night fragrance (Line 4). Since both scents are combined in some individuals (Line 3), the two scents cannot be themselves allelomorphic. It is therefore assumed that each is a distinct inheritance, and that the alternative to each is an absence of that fragrance. In both groups of F1 hybrids, Recurvus scent is strongly dominant, and Tristis scent strongly recessive. Let us first consider each type of fragrance independently. If Recurvus scent were dependent on a single dominant factor, then three-quarters of the F2 should be Recurvus scented, either alone or in combination with Tristis Scent. The remaining one-quarter should then be Tristis scented or non-fragrant. By similar reasoning, one-quarter should be Tristis scented, alone or in combination with Recurvus, if Tristis scent were the result of a single recessive factor. Neither of these two types of fragrance even approaches these expectations. Each falls far short. This being so, either a more complex inheritance or the presence of modifying factors must be assumed.

A Two-Factor Explanation

Possibly two factors are needed to bring out each scent. If this be assumed for Recurvus scent, and we designate the two factors as R and S, and their allelomorphic non-fragrant factors as r and s, then G. recurvus, being homozygous for the factors for Recurvus scent, would have the formula RRSS; and G. tristis, being assumed to be homozygous for the absence of both R and S, would have the formula rrss. Then the F1 hybrid would carry the formula RrSs. Since the F1 has a dominant Recurvus scent, then R and S, both heterozygous, are assumed to be enough to produce Recurvus scent. In the F2 generation there would be sixteen possible permutations of R and r, S and s. Of these, RRSS, RRSs, RrsS, RrSS, RrSs, RrsS, rRSS, rRSs and rRsS: that is, 9 contain at least R and S, and 9/16 or 56¼% should carry Recurvus scent. Actually, combining Recurvus scented F2 (Table I, line 2) with Recurvus and Tristis scented F2 (Line 3) the total Recurvus scented are 196 of an F2 total of 362, or 54.1%. Thus the deviation from the expected value is slightly over 2%, which is regarded as sufficiently close. In the backcross, since RrSs (F1) is crossed with rrss (Tristis II), only 4/16 or 25% would have both R and S, and therefore have Recurvus scent. Actually, combining the values in line 2 and 3, Table I, in the backcross, 70 of 284 total, or 24.7%, are Recurvus scented, which is only 0.3% deviation from the expected 25%!

To get an equally close approximation for the inheritance of Tristis scent, we again assume two necessary factors, which we may designate T and U, and assume that a minimum of heterozygous T and homozygous U are necessary for Tristis scent in the absence of Recurvus scent. But in the presence of Recurvus scent also, then homozygous T as well as homozygous U are needed, with heterozygous or homozygous R and S, to bring out both scents together. Thus Recurvus scent is regarded as epistatic to a minimum dosage for Tristis scent. On this hypothesis, there are now eight factors to consider; R, r, S, s, T, t, U, and u. The possible groupings of these in the F2 are now 256.

Of these, plants having both Tristis and Recurvus scents will then have one of the following genetic formulae in the following proportions: 1 RRSSTTUU, 2 RRSsTTUU, 3 RrSSTTUU, 4 RrSsTTUU, and will constitute 9/256 or 3½% of the total in the F2; this expected value comparing favorably with the actual 2.2% found. In the backcross 1/16 or 6¼% would be expected, and there were actually 6%.

Figure 2
Figure 3
The Tristis X Recurvus F1 hybrids varied in form, color and markings, but not in fragrance. All three of these had the violet scent of G. recurvus. A backcross of the F1 to G. tristis concolor. This particular plant carried both types of fragrance, violet by day and night-jasmine at night.

Individuals having formulas containing TTUU or TtUU, without R and S both being present also, will have only Tristis scent; and will be expected to appear in 21/256 or F2, or 8¼%, and in the backcross 6/16 or 37½%. These expected values are considerably higher than the actual one for Tristis scented alone (Lines 4 and 5). But if we lump together all of the fragrant groups not containing Recurvus scent (that is, all of lines 4 to 8 inclusive), we then closely approach the expected values for Tristis scent alone. The actual values (given as total % under F2 and backcross in Table I) are 8.6% compared to an expected 8¼% in the F2, and 36.2% compared to an expected 37½% in the backcross. We are accordingly led to conclude that all of these: Lemon, Pungent, and Tristis scented and combinations among them, all have the basic genetic formula for Tristis scent, TTUU or TtUU, with possibly other factors as modifiers. The close agreement of the actual proportions of the inheritance of the different types of fragrance with the expected values based on this assumed hypothesis, is reasonable evidence in support of it. This is further corroborated by the few instances of crosses between F1 and garden gladiolus sorts, and by later generations of the fragrant race, in all of which the proportions of fragrances found agreed very well with expectations.

The next practical question is: do the traits of the species, G. tristis and G. recurvus, link strongly enough to the factors for scent to make any of these characters practically unavoidable in the fragrant garden hybrids? Only one species trait seems to be closely enough linked to fragrance to be a real probability. G. recurvus has a grayish leaf color, and Recurvus scent appears to be linked to this 81.4%. None of the other characters that differ in the two species show any high degree of linkage to either type of fragrance. In fact, many of the traits of the Tristis-recurvus hybrids appear to be easily lost in crosses with garden varieties. Thus the nondescript greenish cream colors of these hybrids are replaced by shades of pink and rose in crosses with rose colored garden varieties; by lavender-violet in crosses with colored varieties, and so forth. It thus seems probable that all colors and many of the types found in the garden gladiolus may be obtained in fragrant hybrids.

Hybrids With Gladiolus Sterile

But there is one real drawback. The hybrids between the small fragrant type of sweetglad and the garden varieties are thus far consistently sterile. One reason for this may be in their chromosome numbers. They are all triploid. The little fragrant species are all diploid, with a 2n number of 30, according to Bamford.3 The large flowered forms, according to the same authority, are tetraploid with 60 chromosomes in each body cell. Accordingly, the hybrids combining these two groups have 45 chromosomes, 15 from the fragrant parent and 30 from the large garden variety. Yet even when outnumbered two to one by chromosomes carrying factors for non-fragrance, the two chromosomes of the fifteen, carrying the factors R and S for Recurvus scent, impart their fragrance to the hybrids. The Tristis night scent, requiring one of its two supplementary factors T and U, to be homozygous in order to bring out fragrance, of course is not transmitted to these mixed hybrids. It would require two chromosomes carrying U, in addition to one or two carrying T to transmit Tristis fragrance, and from a diploid parent only one of each of these chromosomes would normally be bequeathed to a hybrid with a non-fragrant variety.

But there are possibilities even here. Since the large flowered gladiolus varieties are tetraploid, then tetraploids may be possible in the small fragrant group, too. Perhaps following the technique of Blakeslee and Avery,3 using colchicine, a tetraploid race of the small fragrant gladiolus may be attained. If this were homozygous for the factors for Tristis scent, then the tetraploid would have in its body cells not two genes each for T and U on four chromosomes, but four chromosomes carrying T and four carrying U. When crossed with a large-flowered tetraploid gladiolus, this fragrant tetraploid would transmit two chromosomes carrying T and two carrying U to the hybrids. Would these tetraploid hybrids, each having two chromosomes each for T and for U, then carry the Tristis scent or not? It is interesting to speculate.

Meanwhile the dominance of the factors R and S for Recurvus scent make it easy to transmit this scent to triploid hybrids with the garden varieties, and the absence of any close linkage of the primitive species traits to the pairs of factors for Recurvus scent, except in the case of gray leaf color, makes it also relatively easy to duplicate in fragrant hybrids any of the color or form characters of the garden gladiolus. True, the triploids are, as might be expected, smaller and more slender than the sturdy garden giants. They thus partake more of the traits of the primulinus race than of the large exhibition type, but without the hooded flower or wide spacing of the primulinus. Not everyone wants giant gladioli, so this daintier type will have a following.

The little fragrant diploid race developed in this study, when crossed with the garden varieties, acts as a cameleon in heredity, taking on more of the characters of the tetraploid large parent than of its own wild ancestors, except in the trait for fragrance, which carries over readily enough. May this mean future fragrant gladioli made to order? It may be even so.

  1. From a study of hybrids between one clon of G. tristis and two of G. recurvus, in the F1, F2, and a backcross of F1 x tristis, the violet-like fragrance of G. recurvus was found to be dominant, and the jasmine-like night fragrance of G. tristis was recessive.
  2. The Recurvus fragrance is dependent on two factors, either or both of which may be heterozygous.
  3. The Tristis fragrance is also dependent on two factors, one of which must be homozygous, while the other may be heterozygous, providing the Recurvus fragrance is not also present. In the latter case it suppresses the minimum dosage of Tristis fragrance, but not a completely homozygous Tristis fragrance.

Literature Cited

  1. BAMFORD, ROLAND. The chromosome number in Gladiolus. Jour. Agr. Res. 51: 945-950. 1935.
  2. BLAKESLEE, ALBERT F. Demonstration of differences between people in the sense of smell. Scientific Monthly 41: 72-84. 1935.
  3. BLAKESLEE, A. F. and A. G. AVERY, Methods of inducing chromosome doubling in plants by treatment with colchicines. Science 86: 408. 1937. Journal of Heredity 28: — 1937.
  4. HERBERT, WM. Instructions for the treatment of Amaryllis longifolia as a hardy aquatic, with some observations on the production of hybrid plants, etc. Trans. Hort. Soc. London 3: 73-80. 1820.
  5. MCLEAN, F. T. The value of Gladiolus species for hybridizing. Glad. Rev. 2(5): 7-9, No. 5, May, 1925.
  6. ————— Inheritance and chromosome number in the Gladiolus. Torreya 31: 65-70. 1931.
  7. ————— The inheritance of fragrance in Gladiolus species crosses. Jour. N. Y. Bot. Gar. 34: 73-80. 1933.
  8. ————— A genetic analysis of the inheritance of fragrance of Gladiolus. Bull. Torrey Bot. Club 65: 181-197. 1938.