This appeared as an article in ALOE 38:31 (2001). Unfortunately there was a problem with illustrations and captions and these are corrected here. A comment is also added as an addendum to respond to criticism by I. Breuer published in Alsterworthia 1:13(2002).

Introduction:

I described Haworthia serrata in 1973 (Jl S.AFr.Bot.39:249, see Figs.1) from Oudekraal, southwest of Heidelberg. I commented then on the wisdom of describing a new species when “the recognition, estimation of taxonomic rank and circumscription of elements in Haworthia…” was so problematic. The new species was said to resemble H. emelyae var. multifolia (Figs.2). In respect of its distribution, I said it was closest to H. heidelbergensis at Heidelberg (Figs.3 JDV87/1) and as at Matjestoon (Fig.4 JDV87/3), and also to H. sublimpidula at Swellendam (now known to be H. floribunda var. major (Fig.5 MBB6859, taxonomically with little connection to H. rossouwii). The implication was that it could have been taxonomically related to those elements in terms of geographic distribution. I was still puzzled by the relationships of H. serrata when I wrote (New Haworthia Handbook :55, 1982) that collections by C. Burgers from the coastal limestones might throw more light on the matter (Fig.6 MBB6985 H. mirabilis var. calcarea).

In his revision of Haworthia (1983), C.L. Scott recognised my species. He confused it, and illustrated it, with a photograph of H. mucronata var. mucronata sensu Bayer (1999), which he called H. mclarenii. This does exemplify the problem of look-alikes in the subgenus, and this curiously does have some connection to the actual problem.

When I wrote Haworthia Revisited (1999), I was still not sure what the real affinities of H. serrata were. I mentioned that my 1982 observations were vague because of the unusual elements which occurred in the coastal limestones. I described these as new: H. mirabilis var. calcarea from De Hoop Nature Reserve (see Fig.6), H. variegata var. petrophila (CBurgers2158) from Karsriver and H. heidelbergensis var. minor from N Bredasdorp (Figs.7 JDV86/5).  In my 1999 classification, I placed H. magnifica var. paradoxa (Figs.8 JDV86/78 & 94/89) in H. mirabilis, suggesting that it may have a connection there via the var. calcarea and an element from the Potberg (CBurgers 2018) which I have not yet seen in the field myself (since I wrote the article I have in fact seen this population, MBB7356, and it is discussed in the Chapter re H. mirabilis). The connections of paradoxa must be viewed in its relationship to H. emelyae var. major, in the same way that serrata is compared with H. emelyae var. multifolia. I also suggested that H. serrata had a very close resemblance to some forms of H. mirabilis var. sublineata from south of Bredasdorp. This present paper addresses some of these observations again.

New Records
In June 1999, I revisited H. serrata at Oudekraalkop (MBB166) again, this time together with J.D.Venter. Despite rumours of over-collection, it was still very abundant, although localised. It was also still present at Koppies (Fig.9 JDV99/6) a short distance to the west, but in very low numbers at a badly burnt site. P.V. Bruyns (priv. comm.) has reported it at a more northerly site as well. This could be where J.N.O. Uys also showed it to me at a second site on Oudekraal when he first introduced me to this species. It struck me again how similar the plants were to H. mirabilis (particularly the var. sublineata, Figs.10 MBB6639, bearing in mind the vast variation that exists there) and to H. heidelbergensis var. minor. The visit in June ’99 was a general excursion to explore the reported presence of H. maraisii at Koppies (ex E. Esterhuizen), in relation to that species and heidelbergensis as I know it in that general area. Our finds and observations led us later in the year to explore the area north of the Potberg. The extraordinary finds there led me to further explore westward to Bredasdorp itself and to see for myself the var. calcarea in the De Hoop. Two visits for that purpose were fruitless, but thanks to Dennis DeKok, I did locate it in the company of Lawrence Loucka in July 2000 (MBB6985-fig.11). This latter excursion was particularly fruitful because, by great good fortune, we discovered H. serrata in unmistakable form north of Bredasdorp (Soutkloof, Fig.12a & b MBB6983), and Nooitgedacht, MBB6984). Both sites are west of Rooivlei where heidelbergensis var. minor occurs. We also collected variegata var. petrophila (Fig. 13 MBB6986) west of the Karsriver limeworks. For that variety I have C. Burger’s record (CB2158) from east of the limeworks, and a collection I made at Renosterfontein (Fig.14 MBB6632) on the east bank of the Karsriver in 1997. This latter collection, in particular, reminded me in both form and colour of H. emelyae var. multifolia (see fig.2).

The conclusion from seeing H. serrata north of Bredasdorp, and both var. calcarea and var. petrophila, is that this is the prime connection. It seems now more improbable that the var. petrophila is as close to H. variegata as I had supposed, and that the var. calcarea is closely connected to both H. serrata and to var. petrophila. The other observations remain valid viz. that there is a strong resemblance between serrata and mirabilis var. sublineata; but particularly that serrata may be connected to heidelbergensis via the var. minor.  However, heidelbergensis in its general relationships is extremely complex. It is interfused with H. maraisii, it may be the direct (if complex) geographic extension of H. mirabilis in the context of the relationship between that species and maraisii, and it may have affinities and continuity with both H. variegata and H. floribunda. There is still another problem: H. variegata var. modesta at Kathoek (Figs.15 JDV97/24) and Potberg (Figs.16 MBB6542).  My collection of this element at Kathoek (Figs.17 MBB2551) suggests that it is possibly a small very proliferous sand dwelling serrata. J.D. Venter’s collection from the same farm is a more robust element with rather longer, more erect leaves which suggests H. variegata.  My Potberg collection (fig.16) is another small element which I relate to variegata although the leaves are shorter and more turgid. It must be borne in mind that H. variegata var. hemicrypta occurs north of the Potberg on the gravelly flat so that there is ecotypic variation in evidence. Further work will have to be done to elucidate what could be an unresolvable situation i.e. the Potberg is home to variants which link all the elements so far named. The evidence for this is already available, but insufficient to crystallise a solution.

There will be difficulty in considering the extent and the degree of similarity of H. serrata to other populations. Some of these which include serrata-look-alikes are:

Some of these populations and their variants are dealt with in other chapters (Particular note must be made of H. elizeae which has to be considered in relation to H. rossouwii).

H. serrata was coming into bud in July a good five months after mirabilis had started to flower. This shows that my predictive thoughts about the relationship with calcarea and petrophila may not be particularly sound if flowering time is considered. Calcarea flowered a little before mirabilis var. sublineata in January, petrophila flowered in February while Roovlei heidelbergensis and Leeurivier (Drew) heidelbergensis flowered in December and March respectively. Flowering time for other collections needs to be observed, noting that this is not necessarily definitive either (for example while H. mirabilis var. badia is essentially late-summer flowering, single plants may flower in winter.

When I consider how much time I have spent in the area, it was really surprising to find serrata so far west.  It means that something very interesting can still be found in the large triangle Swellendam/Bredasdorp/Riviersonderend, and that exploration will have to be done on a very small scale. The Karsriver and Renosterfontein petrophila are just smaller more spinose versions of Oudekraal, and the leaves are more spotted on the reverse. Calcarea was previously known from two clones which had rather shorter leaves and, perhaps, fewer leaves per rosette. This, the locality, and the spotting on the back of the leaves would have been the reason for opting for a relationship with mirabilis. We only found it with quite detailed directions because it was such a small population and so cryptic, consisting of three tiny locations under bush with a total of about 15 clumps. One location could have been the product of vegetative propagation with six to seven clumps. The second was similar with only two clumps and evidence of porcupine activity around the clumps. The third covered about 15sqm and there were another 15 or so plants as clumps or as single rosettes. These did vary and only one clump really resembles the two I have seen in cultivation viz. the Karoo Garden plant of C. Burgers, and a plant I received from Dennis DeKok.  Burgers, however, appears to have made two collections from De Hoop; this needs to be confirmed.  The plants seen by me in habitat do vary quite a bit and the range of variation equates that of both serrata ex Oudekraal (the Koppies population is reduced to about five plants) and north of Bredasdorp.  The mean length of the leaves and leaf number per rosette may be shifted to the shorter and fewer, respectively.

The two Bredasdorp populations of H. serrata are remarkable for the similarity of the plants to serrata at the type locality. The habitat is a little different and at the westernmost of the populations, the plants cluster much more freely than the Heidelberg populations. At Koppies the plants are on a Witteberg substrate, whereas at Oudekraal and N Bredasdorp, it is Bokkeveld shale. We may have a slightly biased image of serrata with turned out leaves; the plants can be quite globose, and I did illustrate a plant like that in my original description when I drew the comparison with herbacea.

Where serrata really differs from mirabilis is by the greater number of leaves per rosette, the thinness of the leaves, narrowness at their bases, lack of a mid‑leaf swelling and by the incurving of the leaf tip; that character is fairly significant.  H. mirabilis var. sublineata grown recently from seed has proved to be remarkably variable, but none of the plants have incurving leaves. Some of the clones are very similar to serrata, but the flowering time is earlier in the year and more in keeping with H. mirabilis in that respect.

Discussion
H. variegata var. petrophila and H. mirabilis var. calcarea should be transferred to the species H. serrata. The same may be true for H. heidelbergensis var. minor. However, H. heidelbergensis has a complex distribution and there are other complexities in variation that preclude a definitive statement. It is quite probable in the light of the findings discussed in Haworthia Update Vol.1, that there is no discontinuity which satisfactorily and convincingly can be claimed to distinguish species. This is probably just another example of a more general problem which I have experienced with the identification and classification of many plant genera.

I am conscious that I have not dealt fully with the problem in omitting mention of co-occurrence (I mean sympatry – when two obviously different elements grow together at the same site or very closely adjacent). I see this as the best measure of discontinuity. Thus if two elements grow in close proximity, even if not sharing the same habitat, they are probably discrete. This does not preclude the possibility that they are simply ecotypes and this is taken into consideration. However, I have before pointed out that a taxonomic solution which can be deduced in this way for populations might not be true for the entire distribution of the species concerned. It is evident that serrata generally occurs alone. At Oudekraal and at Koppies, there is a floribunda/maraisii element in the vicinity.  In the case of heidelbergensis var minor, both turgida and maraisii are present. I have always been struck by the similarity of mirabilis and serrata, and by the fact that they do not co-occur.  Mirabilis, in the context of this paper, is presently not known east of Napier except at Bredasdorp and Mierkraal south of Bredasdorp. This is as the var. sublineata and var. mirabilis respectively, which differ from serrata in respect of flowering time, less leaves per rosette, and the non-incurving leaves.  Curiously, it appears that mutica, at least in the Bredasdorp area, grows generally alone too, although at Haarwegskloof (Bredasdorp) it does grow with heidelbergensis.  At Drew it is in the close company of both maraisii and heidelbergensis, and south of Stormsvlei is near to a maraisii/mirabilis intermediate. Thus even an element as apparently different as H. mutica may presumed to be, cannot be ignored.

However, another consideration now enters the picture with the discovery of serrata so far west.  There are three poorly understood elements which were reported to have come from Bredasdorp and Napier respectively. The first of these was H. bijliana von Poelln. that does, however, seem to belong rather with H. arachnoidea than with anything else. G.G. Smith appears to have had a plant of this ilk which he received from Maj. Venter 9, from Caledon. The other two, however, are H. altilinea var. bicarinata Triebner and H. rossouwii v.Poelln.  They were both described from the same collection viz. Napier at Bredasdorp, Rossouw in Triebner 1059. The former was never illustrated and the description is rather brief … “Backs of leaves always with two keels and with teeth. Leaves 3-5cm long, 6-8mm broad, with 4-6 dark lines in lighter part, with two keels at back, keels with up to 2mm long denticula”.  This description is in the joint paper by Triebner and von Poellnitz, but authorship is attributed to Triebner alone. von Poellnitz later retained this variety when replacing the name altilinea with mucronata (note that in the context of the names which I use, Scott confused my serrata with mucronata as well). This is very curious because it is evident from the description of H. rossouwii, which is very much more detailed, that another plant from the same collection is in fact again described. The differences are essentially that von Poellnitz gives the leaves as 5-6cm long, and 8-10mm wide as opposed to the respective 3-5cm and 6-8mm by Triebner; and with three lines on the ‘end surface’ as opposed to four to six by Triebner. This says a great deal about the detailed descriptions which are routinely called for, and which in Haworthia seldom amount to much because of the problem of look-alikes, and the variability within and between populations.

I repeat the von Poellnitz description here to illustrate that it is a very good description that nevertheless omits some key information:

H. rossouwii von Poellnitz spec.nov. (Stemless rosette, spirally many-leaved. Leaves rather pale green, nearly oblong lanceolate, acuminate, on the face somewhat retused near the apex thus forming a triangular face; on the back, towards the apex, with sub-pellucid tubercle-like spots, denticulate on the margins and keels; the triangular face strongly acuminate, sub-translucent, not tubercled, but carrying some sub-pellucid small teeth, and traversed by a few longitudinal lines).

Stemless many-leaved rosette about 5-6cm in diameter.  Leaves more or less erect, 5-6cm long, 8-10mm wide, about oblong-lanceolate, tapering gradually, pointed, slightly widened at the base, almost straight or a little bent to one side near the tip, rather light green, dull.  On the face flat or usually somewhat convex, near the tip slightly bent backwards (retused) at an angle of 20 degrees; on the back, from the middle outward or near the tip obliquely keeled, or somewhat more rarely with two keels, near the tip with a few, often elongated, very often raised slightly pellucid spots, and frequently with a few leaf-coloured tubercles.  The end-surface half-pellucid, 20mm long or somewhat longer, 7-9mm wide at the base, triangular, tapering very gradually, pointed at the tip often slightly curved inward, somewhat convex, smooth, rarely with a few indications of tubercles, but with some semi-translucent teeth (these teeth loosely arranged in a lengthwise middle line, now and then also in another line nearer the margin and very rarely scattered about in small numbers) with three long, occasionally interrupted, green lengthwise lines, of which only the middle one may reach almost to the apex, and usually with a few very short line; leaf margins and keel with pale teeth, or near the base often leaf-coloured, teeth, up to barely 2mm long; end-bristle pale, with small side bristles, 5-7mm long.

… On account of the teeth on the end surface stands near to H. schuldtiana and to H. paradoxa (I treat these respectively as H. maraisii and H. mirabilis var. paradoxa) but is easily distinguished from both. Its flowers are thus far unknown”.

The omissions are the flower of course, and there is no stipulation of the approximate number of leaves, and leaf thickness. The weakness is that it is a single plant description, evidenced by the fact that another plant from the same collection is accepted as a variant of a totally different complex.

In the past I have I regarded rossouwii speculatively as a synonym of H. mirabilis, due largely to the fact that it was recorded from the Bredasdorp area (and to the ambiguity of the type). Another reason was the similarity I observed between the illustration of H. rossouwii in Kakteenkunde  with H. heidelbergensis var. minor and with H. mirabilis var sublineata. I opted for the latter (with which Scott was in accord) and the similarities are also recorded in the discussion above. I dismissed bicarinata as insufficiently known because I assumed that Triebner may have had some better reason to have equated his specimen with the Little Karoo H. mucronata. Possibly there has been a mistake in the Triebner numbers. It is this discovery of serrata so close to Bredasdorp, and equidistant from Napier, which revives and also casts new light on the problem. I conclude that my H. serrata is a later synonym of H. rossouwii of von Poellnitz.

There are still two other considerations. In my 1999 revision, the fact is that I placed paradoxa as a variety of mirabilis while discussing the possible connection of these elements via calcarea.  In the light of this paper it is evident to me that paradoxa could also be linked to serrata rather than with mirabilis. The same is true for emelyae var. multifolia. I would not like to commit myself to this probable solution at this stage, as I think more evidence should be obtained. It is also improbable that, however attractive, the solution will not avoid impacting on H. emelyae, H. magnifica and H. heidelbergensis. It is also evident that the requirements of the ICBN are that should paradoxa indeed need to be incorporated into serrata, more name changes will result, as paradoxa will have priority over rossouwii.  My opinion is that this is a fundamental weakness in the code, which inhibits speculative treatments, and also which results in smaller taxa becoming the typical main element of a species. My further opinion is that it is a weakness of the Code that the description of an element automatically creates an equivalent taxon. This may be acceptable and desirable where the elements are clearly separable and geographically defined, but not where the typical element is required to absorb any other variants which are not so defined. Thus H. paradoxa var. paradoxa could become the main component of a species where it is actually a geographical isolate. In a system which does not have an adequate general definition for ‘species’, this probably does not matter; whereas it might if there is a definition which suggests that the ‘species’ is an operating living system (see Bayer in Asklepios 77:3, 1999).

The formal statement of this paper, which would better be regarded as an hypothesis subject to test, is thus as follows:

H. rossouwii von Poelln. in Kakteenk.7:75 (1938).  Type: Napier near Bredasdorp, collected by Mr Rossouw, South Africa Police in Napier, Triebner 1059. Not preserved.  Lectotype: designated by Breuer & Metzing 1998, icon in Kakteenk.  non syn. H. mirabilis sensu Scott 1985, Bayer 1976, 1982, 1999.

H. altilinea var. bicarinata Triebn. in Triebner & von Poellnitz in Feddes Repert.Spec.Nov. 45:170 (1938).  Type: Napier at Bredasdorp leg. Mr Rossouw (=Triebn.1059).  Not preserved.

H. serrata Bayer, Jl S.Afr.Bot. 39:249(1973).  Bayer :55(1976).  Bayer :147(1999).  non Scott :62(1985).  Type: CAPE‑3420(Bredasdorp): Oudekraalkop, Heidelberg (‑BA), Bayer 166 (NBG).

Material seen:- 3420(Bredasdorp): Soutkloof, Bredasdorp (-AC), Bayer 6983;  Nooitgedacht, Bredasdorp (-AC), Bayer 6984;  Oudekraalkop, Heidelberg (‑BA), Bayer 166 (NBG); Koppies (-BA), Bayer 4902 (NBG); N. Oudekraalkop (-BA), Bruyns 6260 (NBG).

H. rossouwii var. calcarea (Bayer) comb.nov.  H. mirabilis var. calcarea Bayer in Haworthia Revisited :110(1999).  Type: CAPE-3420(Bredasdorp): De Hoop (-AD), C. Burgers 1648 (NBG, Holo.).

Material seen:- 3420(Bredasdorp): NNE. Buffelsfontein, Potberg (-BC), Burgers 2018 (NBG).  De Hoop (-AD), C. Burgers 1648 (NBG), Bayer 6985.

H. rossouwii var. petrophila (Bayer) comb.nov.  H. variegata var. petrophila Bayer in Haworthia Revisited:159(1999).  Type: CAPE-3420(Bredasdorp): Renosterfontein (-AC), Burgers 2158 (NBG, Holo.).

Material seen:- 3420(Bredasdorp): Renosterfontein (-AC), Burgers 2158 (NBG), (-CA), Bayer 6632;  Karsriver (-AC) Bayer 6986;

The provisions of this solution are that the epithet paradoxa may be instated preferentially, and that emelyae var. multifolia and variegata var. modesta also be included.

A second provision is that the species H. bijliana v.Poelln. be rejected as insufficiently described and sourced, and that the type (the illustration in the Berlin herbarium) is ambiguous – as indeed are many Haworthia types and descriptions based on single plants and not correctly tied to geographical origin. This is demonstrated by the citation by VPoelln. (Feddes Rep. 1936) of this species viz. bijliana from Klawer and Eenriet and also Feddes Rep. in 1937 from Springbok – which cannot be anything other than H. arachnoidea. In this last publication he cites a specimen from Bonnievale, another from Ezeljacht Oudtshoorn and also sinks fergusoniae here which in fact was supposed to have come from ‘near’ Grahamstown.

H. bijliana Poelln.  in Feddes Repert. 27:134 (1929); Bredasdorp, Mrs va der Bijl.  Type van derBijl s.n. in (B). Not preserved.  Lectotype desig. Breuer 1999, unpublished photo. icon. (B).  Species rejecienda.

Acknowledgement
I am grateful for the company on the field trip of Lawrence Loucka of Connecticut (USA), and that of my wife, Daphne. The assistance of Paul Botma of Cape Nature Conservation is acknowledged over and above the general acknowledgement to that Department. Mr. C.L. Neethling, Mr. Thys Swart and Mr. Pieter deKok were most kind and helpful in respect of access to sites. I am also grateful to Kobus Venter for support, advice, and constructive exchange of ideas. Not least to Steven Hammer, with comment from Bob Kent, who was so kind as to read this manuscript critically and make corrections. Residual errors will be mine.

Addendum
Alsterworthia Vol 1:13(2002) “A comment on Bayer’s latest taxonomic contribution” by Ingo Breuer. [link] With reference to the comments in Alsterworthia 1:13(2002) regarding the identity of H. rossouwii and H. serrata:

Notes

Fig. 27 A copy of the von Poellnitz lectotype of H. rossouwii.

Figs 28a & b Black and white profiles from two angles, of the same one plant actively growing, undessicated and turgid. The plant is a clone from Bayer 6984 Nooitgedacht, NW Bredasdorp. This is correctly H. rossouwii, and correctly H. serrata.

Figs 29a & b. The colour originals of the preceding.

Fig.30 A second plant of MBB6984. This is a more vigorous grower and it shows the differences that obviously confound the uninitiated. There were not many plants at Nooitgedacht. I only collected from two clones and this is a photograph of the second.

Fig.31 A photograph of a plant of MBB6983 Soutkloof, north of Bredasdorp. Showing a plant in which the convexity of the mid-upper surface (and concavity of the lower surface) is virtually absent. I could have, from this sample (MBB6983), produced pictures to match Figs.27 to 29. So no one should claim that Fig.30 is a different species as someone now well might.

Figs 32a & b. Scanned and manipulated copies of figs 28a & b.

Fig.33 Flowers of MBB6672 H. maraisii Koningsrivier, SW Robertson.

Figs.7a, b & c.  These are clones of JDV86/5 H. heidelbergensis var. minor, Rooivlei, NNE Bredasdorp. This is a small population, of also small plants, from northeast of the collection MBB6984, Nooitgedacht. The plants partly resemble H. rossouwii from there, and partly the var. petrophila from Karsriver. Classifying them as a variety of H. heidelbergensis is not necessarily correct. Considering habitat and substrate, one must consider that ecotypification is playing a role, and that was habitat available, there would be further populations which would confirm continuity with these known collections.

Regarding Figs.28a & b.  A bit of play with light and focus and the keel and under-leaf spotting could be varied.  I would like to suggest that they be considered them faded, less sharply focused, printed with less contrast and on a coarser grained paper. Figs.32a & 32b are in fact such scanned copies of Figs.29a & b. I do not think it even necessary to consider that the original rossouwii of Fig.1 endured a long journey in the post to a foreign land. The assertion by Breuer in Alsterworthia is that not only are the plants I figured as H. rossouwii different from Fig.1, but that this Fig.1 lectotype is “probably” of the maraisii-complex. This is at variance with…

1. the Triebner description of the same collection as a variety of H. mucronata;

2. with Col. Scott who regarded it as H. mirabilis. What to say of von Poellnitz’ own assessment – the man who described both species i.e. maraisii and rossouwii in the first instance.

3. with the facts of the pictures before you.

What should an observer make of maraisii growing very near to rossouwii at Oudekraal and near to rossouwii at Napier?  Can we really see a closer resemblance of Figs 27 & 28 with those long attenuate, double-curved, thin leaves – to H. maraisii rather than to H. serrata. I truly do not see that such a judgement can be seriously considered in the light of Breuer’s interpretations. It is of course valid when one considers all the variants that I illustrated in my article above and in terms of my over all classification.

Generally these plants of the SW Cape tend to have the double curvature of the leaves as is evident in the VonPoellnitz lectotype, but one can find in the same population and/or taxon, clones with incurved, outcurved or erect leaves (see fig.30). Broadly speaking H. maraisii has thick recurved leaves. But this is a non-argument even if one (outside of a “lumping vs. splitting” red-herring) considers the complexity of the so-called maraisii-complex, and I am sure that Breuer does not. I frankly think that most observers cannot even dream how complex it is and that it is just braggadocio to write as though anyone can. Breuer implies that one can make the inference of similarity between rossouwii and maraisii from the illustration (Fig. 27), the description and its given locality. This is patently and self-evidently false as is evident from the fact that I used this same information to reasonably deduce the similarity of rossouwii and serrata.

I did not originally submit a copy of the von Poellnitz lectotype, perhaps because I felt that the response “you will see that rossouwii is quite a different plant from serrata“, is what I could have expected from an uninformed audience! The fact is that all the evidence, circumstantial and descriptive, show that they are the same. Including the lectotype would have just been “overkill” and I have never before felt required to write for a readership with less than common sense or basic knowledge, nor for obstructive readers.

It should not be necessary to say this, but the character of the flowers cannot be specified in the way that writers are inclined to suggest. They have this same enormous range of variability and it is just fraudulent to say that there are floral characters by which these taxa (the varieties of H. rossouwii which I now recognise) can be unmistakeably differentiated.  It is the typological concept which drives people to think there is a single description which can encompass the description of a taxon. This is nonsense. It is a statistical problem and we have the difficulty of establishing the probabilities associated with “decision”. Breuer writes “… von Poellnitz’ description probably depends only on 1 or very few selected plants and no other records are reported for this taxon”. This is circa 1945 circumlocution, it is unnecessary and it is untrue. There was never a question of “selection”, and furthermore we have the Triebner description of the same collection as H. altilinea var. bicarinata. Why this sudden doubt about the interpretation of types. It is a shifting of goalposts.

There are few collections in flower now in April, but I could take this picture of three flowers from the one collection MBB6672 H. maraisii – Koningsriver, SW Robertson (Fig.33). It is not an exceptional case or an exceptional collection. But one can see the differences in thickness of peduncle and variation in the individual flowers. It can happen like this, albeit to lesser degree, on the same plant. Colour is also frequently quite variable between and often within collections. Many years ago I undertook to illustrate all the flowers in my maraisii collections. But they were so variable that I just gave it up. The variability in the flowers mirrors the variability of the rosette and the latter is just a little easier to deal with.  One has the added problem with flowers, of the changes which take place during the opening and aging process. So then how does one define the flower, say, of H. mirabilis. It cannot even be done for the range of leaf rosettes, and this has now been shown in the literature many times.

May I also point out that one of my understandings of botany and plant classification is that there is a tendency to overreach ones talents. I have freely admitted that I am there because of the paucity of talent. I frankly do not think the environment is enriched by the particular article I now refer to in these added notes.

I can point out that there are writers who really persist in flogging dead horses. The result is phantasmagoric fanfaronades. The article concerning the name H. correcta (Alsterworthia 1:6, 2001) is not particularly accurate either and indeed it may be my own fault. Smith explained that Mrs. Blackburn did not in fact collect correcta ‑ it was a Mrs. le Roux and the plants came from near Vanwyksdorp. Later Mrs. Ferguson sent a plant from 26 miles east of the town. Thus from this information H. correcta has to be a later synonym for H. emelyae.

Regarding types:- the painful realisation was actually expressed by Schelpe in the distant past, and brought home to me many times. These old types are doubtful and ambiguous.  I was faced with that option long ago – just forget them all, or try to reach a compromise with the understanding and information available at the time. H. rossouwii is a case where there is real new information which would have made me use the name rossouwii in 1976 and before, had it been available. But here, had I persisted with the name serrata for this new material MBB6983 and 6984, there is absolutely no doubt in my mind that I would have been criticised for failing to see the connection to rossouwii. Not that this would concern me unduly, for I think there is actually very little benefit in this mindless re-interpretation of types which takes us nowhere. Notwithstanding, and without compromising, anything so far written:- the really disconcerting thing about the whole matter is that without the locality data, it will be possible to generate an argument bolstered by illustrations, to show that rossouwii was in fact H. decipiens, H. cooperi or H. herbacea (among others), of the option less improbable than H. maraisii.

The closing paragraphs of the article are a rather hackneyed parody sung by most “wannabe” taxonomists. Of course it is right when Breuer says if we lump it means we would have to lump many other things. It is just as true to say, if we split we have to split many other things. That does not make sense either. This is ancient stuff of 60 years ago and it seems strange that it still provides so much fodder for writers, who have in fact no rational species concept to begin with. The Linnaean system was based very largely on floral characters and hence taxa at all levels are based on “related floral characters”. To suggest that this pertains in some special way to “series” is not correct. This sums up the article in Alsterworthia. ♦