Specimens examined: New Mexico (Wislizenus of 1846; Wright 384, 531, of 1852; G. R. Vasey of 1881): Texas (Wright of 1850, 1851, 1852; Bigelow of 1852): Chihuahua (Evans of 1891; Budd of 1891): also growing in Mo. Bot. Gard. 1893.

This species shows an interesting transition from Coryphantha to Echinocactus. The woolly groove of the Coryphantha extends from the spine-bearing areola to the axil of the tubercle, where it expands into the flower-bearing areola. In C. macromeris the groove extends only about half way down the tubercle and gives origin to the flower-bearing areola on the side of the tubercle; while in Echinocactus the flower-bearing areola becomes adjacent to the spine-bearing areola and the flower appears at the summit of the tubercle.

ARTIFICIAL KEY TO THE SPECIES.

It seems impossible to make a simple artificial key that will serve as a useful guide to each individual species and variety. Our knowledge of so many of the species is imperfect, that no set of characters can be applied throughout. However, as no plants are collected in such fragmentary condition, it will be useful to construct a key based upon such characters as are always likely to be present, even if specific distinctions are not always reached. In many cases, species are so closely and differently related to each other that the complete descriptions will have to be consulted to determine the differences, and in such cases the artificial key can only indicate the group. Even the full descriptions are very compact, all characters not necessary for discrimination having been eliminated. No attempt need be made to determine any species by means of the flowers alone. In most cases more or less of the plant body will be available, presenting spine and tubercle characters, and these are used in the following key. The distinction between Eumamillaria and Coryphantha, on the basis of grooveless and grooved tubercles should always be made out easily. It may be useful to suggest as a caution, however, that often tubercles in drying develop folds which simulate grooves, and especially is this true in quadrangular tubercles. In such cases it is necessary to restore the original plumpness of the tubercle by boiling, before the presence or absence of the groove can be definitely determined. The species and varieties are indicated only by their specific or varietal names in the following key, and the numbers refer to the serial numbers of the synoptical presentation. Forms occurring within the United States are marked with an "*":

I. Tubercles never grooved.

* Central spines none.

Radials 5 to 9, stout. meiacanthus* (7).

Radials 20 to 40. micromeris* (12), greggii (13).

Radials 40 to 80. lasiacanthus* (10), denudatus* (11).

** Central spine solitary and not hooked.

+ Central spine longer than the radials.

Radials 7 or 8: tubercles very long (40 to 50 mm.). longimamma (36).

Radials 15 to 20: tubercles 6 to 8 mm. long. eschanzieri (21).

+ Central spine shorter than the radials.

Radials 5 to 9, stout. meiacanthus* (7).

Radials 9 to 22. heyderi* (5), hemisphaericus* (6), gummiferus (8), gabbii (34), sphaericus (35).

*** Central spine solitary and hooked.

+ Stems slender cylindric: Lower Californian.

Centrals 1, 20 to 30 mm. long. roseanus (23).

Centrals 1 to 4, 20 to 50 mm. long. setispinus (24).

+ Stems depressed-globose to ovate.

Radials 4 to 6, rigid. uncinatus (9).

Radials 8 to 12. wrightii* (15).

Radials 15 to 30. grahami* (19), eschanzieri (21).

Radials 50 to 60. barbatus (18).

**** Central spines more than one, and none of them hooked.

+ Slender or sometimes stout cylindrical plants, branching at base: Lower Californian.

brandegei (3), setispinus (24), halei (25).

+ Depressed-globose to ovate and stout cylindrical.

+ Radials few (3 to 12) and rigid: Mexican.

Radials 3: centrals 3. alternatus (1)

Radials 7 or 8: tubercles 40 to 50 mm. long. longimamma (36).

Radials 10 to 12: tubercles 12 to 15 mm. long. gummiferus (8). + Radials numerous (16 to 60), capillary or bristle-like.

Radials 15 to 30, slender but rigid (bristly). acanthophlegmus(2), densispinus (4), bispinus (14), rhodanthus (26), sulphureospinus (27), palmeri (29), pringlei (32).

Radials 30 to 60 or more, mostly capillary. tetrancistrus* (22), capillaris (28), texanus* (31), spaerotrichus (33).

***** Central spines more than one and but one of them hooked.

Radials 10 to 15. goodrichii* (16), setispinus (24).

Radials 15 to 30. pondii (17), grahami* (19), bocasanus (20).

Radials 30 to 60. tetrancistrus (22).

****** Central spines more than one, and more then one of them hooked.

Radials 8 to 12. wrightii* (15).

Radials 30 to 60. tetrancistrus (22).

II. Tubercles with a more or less prominent groove.

* Central spines none.

+ Radials whitish and rigid, oppressed (pectinate) and interwoven with adjacent clusters.

Depressed-globose and simple. compactus (44).

Globose and simple. radians* (45), corniferus (47).

Cespitose. pectenoides (46), sulcatus* (49).

+ Radials more slender and spreading.

Radials 10 to 17. missouriensis* (37), similis* (38), macromeris* (64).

Radials 30 to 50, capillary. dasyacanthus* (51).

** Central spine solitary, not hooked.

+ Central spine porrect.

Radials 6 to 17. missouriensis* (37), robustior* (39), scheerii* (40).

Radials 30 to 50, white and capillary. dasyacanthus* (51).

+ Central spine curved downwards.

Radials 8 to 12. sulcatus* (49).

Radials 12 to 26. robustispinus (41), recurvatus (42), corniferus (47), scolymoides* (48).

+ Central spine erect: Mexican.

Radials 7 or 8: central 50 mm. long. salm-dyckianus (43).

Radials 10 or 11: central 25 to 35 mm. long. maculatus (52).

Radials 13 to 16. compactus (44).

*** Central spine solitary and hooked.

brunneus (53).

**** Central spines more than one and none of them hooked.

Centrals 2: radials 6 to 20. scheerii* (40), robustispinus (41), recurvatus (42), scolymoides* (48).

Centrals 3: radials 6 to 40. scheerii* (40), scolymoides* (48), echinus* (50), conoideus (54), neo-mexicanus* (59), arizonicus* (60).

Centrals 4 or 5: radials 6 to 40. scheerii* (40), scolymoides* (48), echinus* (50), conoideus (54), tuberculosus* (56), viviparus* (57), radiosus* (58), neo-mexicanus* (59). arizonicus* (60), macromeris* (64).

Centrals 6 or 7: radials 12 to 40. potsii* (55), tuberculosus*(56), viviparus* (57), neo-mexicanus* (59), arizonicus* (60), chloranthus (62).

Centrals 8 to 14: radials 12 to 40 or more. potsii* (55), tuberculosus* (56), viviparus* (57), neo-mexicanus* (59), deserti* (61), chloranthus* (62), alversoni* (63).

GEOGRAPHICAL DISTRIBUTION

It is only possible to deal with the forms that occur within the borders of the United States, as even individual stations of common Mexican forms are little if at all known. These United States forms represent a northern extension of an abundant Mexican display. The group EUMAMILLARIA, containing twelve of the thirty-one forms defined as occurring north of the Rio Grande, makes the feeblest extension northward, at no place being found far from the boundary, and all the twelve are Mexican forms which extend but slightly into the United States. Only five of the forms are found east of the Pecos: heyderi, the most widely distributed EUMAMILLARIA, extending from the southeastern border of Texas westward along the whole Mexican boundary except in California; hemisphaericus, extending through southern Texas and southern New Mexico; meiacanthus, also along the Mexican border of Texas and New Mexico; texanus, a low ground form of the Rio Grande Valley, extending from the mouth of the river to El Paso, and suggesting a connection with the West Indian stellatus; and sphaericus, another low ground valley form of similar range, but apparently only extending up the Rio Grande to the region of Eagle Pass.

The Pecos forms the eastern boundary of five other EUMAMILLARIA forms: micromeris, extending northward from Coahuila and Chihuahua, apparently only in the mountains between the Pecos and El Paso; wrightii, of similar narrow northward extension, but ranging further northward on the high plains of the Upper Pecos in New Mexico; denudatus, also with a narrow northward extension west of the Pecos; lasiacanthus, extending from Chihuahua with a northern limit between the Pecos and Arizona; and grahami, a Sonoran type which has spread between the Pecos and southeastern California.

The ten preceding forms have evidently entered our borders from the highlands of Sonora and Chihuahua, with the exception of the Rio Grande Valley forms, texanus and sphaericus. Another species, tetrancistrus, is also a Sonoran type which has reached the eastern slopes of the mountains of southeastern California, and extended through western Arizona to southern Nevada and southern Utah, the most extended northern range of any EUMAMILLARIA. The twelfth form, goodrichii, is Lower Californian, and extends into California only in San Diego County. A summarized statement of the distribution of our twelve EUMAMILLARIA would be that two of them have extended from the low grounds of Coahuila and Chihuahua and spread along the valley of the Rio Grande; nine have come from the high grounds of Chihuahua and Sonora, four of which have extended eastward to the low levels of southeastern Texas; four have kept to the highlands west of the Pecos, and one has kept to the Colorado Valley and its tributaries, while one has a short northern extension from Lower California.

The nineteen forms of CORYPHANTHA are decidedly more northern in their distribution, and are our characteristic representatives of the genus Cactus. Ten of these, however, are but northern extensions of Mexican forms, and six of the ten have simply that tongue-like northern extension in the mountains between the Pecos and the Upper Rio Grande (above. El Paso), viz.: dasyacanthus, tuberculosus, scheerii (which has also spread somewhat east of the Pecos), and the three pectinate and closely related forms radians, echinus, and scolymoides. Of the four remaining Mexican forms, macromeris is a low ground Rio Grande Valley form, extending from above El Paso well towards the Lower Rio Grande; potsii just crosses the border in the neighborhood of Laredo; and radiosus and neo-mexicanus have by far the greatest northern extension, stretching from Sonora and Chihuahua to southern Utah and central Colorado, and eastward to the Guadalupe River of Texas.

The nine remaining coryphanths are distinctly forms of the United States, occupying two well-marked regions, viz.: the northern plains, and the desert region of western Arizona and adjacent California, Nevada, and Utah. In the former region is found the widespread viviparus, which extends from the southern borders of British America to the plains of eastern Colorado and western Kansas, and even crosses the Rocky Mountain divide into northern Idaho and northeastern Washington; and missouriensis, which also ranges from the high prairies of the Upper Missouri to the same southern limit, and is continued southward into Texas in its varieties similis and robustior.

In the Arizona desert region, four distinct but closely allied forms have become differentiated from the strong radiosus stock, viz.: arizonicus, deserti, alversoni, and chloranthus, all of which might be regarded as distinct species. In southeastern Texas is found an isolated form, sulcatus, occurring between the Brazos and Nueces rivers. That viviparus must be regarded as a strong northern extension of the radiosus stock can not be doubted, as the low depressed cespitose northern form seems to merge southward so gradually into the simple more robust ovate to cylindrical forms of radiosus as to suggest the propriety of regarding them all as specifically identical.

The result of a closer inspection of the distribution of these nearly related forms is worthy of note. C. viviparus extends from British America and the Upper Missouri to eastern Colorado and western Kansas; neo-mexicanus (the form most nearly related to viviparus) extends from central Colorado and southern Utah into Mexico; at the southeastern edge of this range begins radiosus and extends eastward through southern Texas; from the western edge of neo-mexicanus the form arizonicus extends westward into southern California, touching chloranthus at its Utah limit, and at its California extension reaching alversoni and deserti, the latter of which extends northward into the desert region of southeastern California and adjacent Nevada. Taking this type as of Mexican origin, it seems to have entered the United States from Sonora and Chihuahua, and to have spread in three directions, viz.: eastward through southern Texas; westward and northwestward into southern California and southern, Utah; and northward to the head waters of the Missouri and British America, though we would limit the northern extension of the present specific type to central Colorado, and would regard the still more northern forms as of the same origin but entitled to specific rank.

2. ANHALONIUM Lem. Cact. Gen. Nov. (1839).

Depressed or flattened, simple, unarmed plants, covered with peculiar imbricated tubercles above and their scale-like remains below: tubercle with lower and upper parts very different; lower part comparatively thin and flat; upper exposed part triangular in outline and divergent, very thick and hard, the lower surface smooth and keeled, the upper surface plane or convex, smooth or tuberculate or variously fissured, with a broad wool-bearing groove or simply a more or less evident tomentulose apical areola: spine-bearing areola obsolete: flower-bearing areola at the summit of the lower peduncle-like portion of the very young tubercle (thus appearing axillary with reference to the exposed part of the tubercle) and bearing a dense penicellate tuft of long soft hairs which conceals the lower part of the flower and the entire fruit and persists about the apical region of the plant as matted and apparently axillary wool: ovary naked: seeds large, black, and tuberculate: embryo obovate, straight.

According to the present views concerning generic limitations in Cactaceae, Anhalonium must certainly be kept distinct from Mamillaria, and to such a view Dr. Engelmann had finally come. The generic distinction is based upon such characters as (1) the complete suppression of the spine-bearing areolae; (2) the strong differentiation of the tubercles into two very distinct regions; (3) the production of the flower at the apex of the basal or penduncle-like portion (which becomes flattened and expanded at maturity) of a very young tubercle; and (4) the large tuberculate seeds.

In the case of engelmanni the broad woolly groove of the upper portion of the tubercle expands below into the flower-bearing areola, but terminates blindly above just behind the sharp apex. In prismaticum and furfuraceum the groove is obliterated, but there usually remains a small (more or less tufted) areola and depression just behind the apex to mark its upper extremity. This apical areola therefore, does not represent a spine-bearing areola, but the closed upper extremity of a tubercle groove.

It seems evident that Anhalonium is a much modified Cactus, and that its affinity is with the coryphanths, through such a species as C. macromeris, in which the flower becomes extra-axillary. If in macromeris, with the flower standing well up on the tubercle, the portions of the tubercle above and below the flower should become very different from each other, the upper portion being so much modified as to cause the spine-bearing areola to be obliterated, the condition of things in Anhalonium would be obtained.

* Upper surface of tubercle with a broad and deep wool-bearing longitudinal groove which widens below.

1. Anhalonium engelmanni Lem. Cact 42 (1839).

Mamillaria fissurata Engelm. Syn. Cact. 270 (1856). Anhalonium fissuratum Engelm. Bot. Mex. Bound. 75 (1859).

Depressed globose or flat, top-shaped below and tapering into a thick root, 5 to 12 cm. in diameter: tubercles (upper portion) appressed-imbricate, 12 to 18 mm. long and about as wide at base, the upper surface convex and variously fissured (presenting an irregular warty appearance) even to the edges: flowers apparently central, about 2.5 cm. long and broad, shading from whitish to rose: fruit oval, pale green, about 10 mm. long: seeds 1.6 mm. long. (Ill. Bot. Mex. Bound. t. 16) Type unknown; but specimens of Wright, Bigelow, and Parry in Herb. Mo. Bot. Gard. are the basis of Engelmann's Mamillaria fissurata.

On limestone hills, in the "Great Bend" region of the Rio Grande in Texas, and southward into Coahuila. Fl. September-October.

Specimens examined: Texas (Wright of 1850; Bigelow of 1852; Parry, with no number or date; Lloyd of 1890; Evans of 1891; Briggs of 1892): also growing in Mo. Bot. Gard. 1893.

This species is very closely related to the Mexican A. kotchubeyi Lem. (A. sulcatum Salm-Dyck), but unfortunately no type of that species seems to be in existence, and Dr. Engelmann notes (Mex. Bound. Rep. 75) that "it seems no living or dead specimen is at present extant in Europe." Judging from the description, the upper surface of the tubercles in A. kotchubeyi, aside from the central furrow, is smooth; at least the margin is "very entire."

** Upper surface of tubercle not grooved, but usually with a tomentose pulvillus at the tip.

2. Anhalonium prismaticum Lem. Cact. 1 (1839).

Mamillaria prismatica Lem. Hort. Univ. i. 231 (1839). Cactus prismaticus Kuntze, Rev. Gen. Pl. 261 (1891).

Flat above, top-shaped below, 7.5 to 12.5 cm. in diameter: tubercles (upper portion) close]y imbricate but squarrose- spreading, sharply triangular-pyramidal and very acute (with a sharp cartilaginous tip, which usually disappears with age and leaves the older tubercles blunt or retuse), 18 to 25 mm. long and about as wide at base, the upper surface almost plane and smooth, except that it is more or less pulverulent and usually bears a small tomentose pulvillus (often evanescent later) just behind the claw-like tip: flowers rose-color: fruit elongated- oval and reddish. (Ill. Lem. Cact. t. 1.) Type unknown.

Referred to Mexico in general, but reported definitely only from San Luis Potosi. Undoubtedly found in Coahuila, and possibly crosses the Rio Grande in the region of the "Great Bend."

Specimens examined: San Luis Potosi (Eschanzier of 1891): Mexico in general (specimens from Coll. Salm-Dyck in 1858; Schott of 1858): also specimens cultivated in Mo. Bot. Gard. in 1881; also growing in same garden in 1893.

3. Anhalonium furfuraceum (Watson).

Mamillaria furfuracea Watson, Proc. Amer. Acad. xxv. 150 (1890).

Very closely related to prismaticum; but triangular portion of tubercle acuminate and shorter, having an irregularly mamillate upper surface, and the acumination ending abruptly in a cartilaginous depression containing a tomentose pulvillus: flowers 2.5 to 3 cm. long, white or pinkish, the sepals brownish. Type, Pringle 2580 in Gray Herb.

At Carneros Pass, Coahuila.

Specimens examined: Coahuila (Pringle 2580 of 1889).

The type of this species was not among the collections received from Cambridge, but a specimen of the same distribution from the National Herbarium shows tubercle dimensions different from those recorded in Dr. Watson's description. In that description the triangular terminal surface is said to be "about an inch broad by one-half inch," which is decidedly different from the equilateral surface of the tubercle of prismaticum. In the National Herbarium specimen of furfuraceum, however, of the same distribution, the surface is almost equilateral, measuring 15 mm. long by 18 mm. wide at base. Without the acuminate upper portion the breadth of the triangular portion would be about double its length. The lower rim of the cup-like depression which terminates the tubercle and contains the pulvillus is sometimes slightly prolonged into a tooth, which in prismaticum becomes the sharp tip of the tubercle. The "minutely furfuraceous-punctulate" character of the tubercle is common to all the species of Anhalonium I have seen, and simply represents the external openings of the remarkably long cuticular passageways to the stomata.

4. Anhalonium pulvilligerum Lem. Cact. (1839).

Anhalonium elongatum Salm-Dyck (1850).

This seems to be a third grooveless Mexican species. I have seen no specimens, but judge from the description that it differs from the two preceding species chiefly in its less crowded and more elongated tubercles (triangular portion 5 cm. long by 2.5 cm. broad at base), which are covered at apex with a tomentose pulvillus.

GEOGRAPHICAL DISTRIBUTION.

This curious genus is strictly Mexican, and, so far as at present recorded, is characteristic of Coahuila, but a single species (engelmanni) of the four or five known crossing the Rio Grande in the Great Bend.

3. LOPHOPHORA, gen. nov.

Depressed-globose, proliferous and cespitose, tuberculate-ribbed, unarmed plants: tubercles at first conical and bearing at summit a flower-bearing areola with a dense tuft or short pencil of compact erect hairs, when mature becoming broad and rounded (with the remnant of the penicellate tuft as a persistent pulvillus in a small central depression) and coalescing into broad convex vertical ribs: spine bearing areolae obsolete: flowers borne at the summit of nascent tubercles: ovary naked (that is free from scales, but often downy): fruit and seed unknown.

These forms have been variously referred to Anhalonium and Echinocactus, but seem to deserve generic distinction. They differ from Anhalonium in the entire suppression of the upper highly differentiated portion of the tubercle, in the broad and rounded development of the lower portion, and in the coalescence of the enlarged tubercles into broad vertical ribs. In fact, in young specimens, the plant appears almost smooth, with shallow furrows radiating from the depressed apex. The genus differs from Echinocactus in the suppression of the spine-bearing areolae, and the naked ovary. In the examination of developing tubercles the relation to Anhalonium is evident. In the latter genus the young tubercle bears on the summit of its pedicel-like lower portion the tufted flower-bearing areola the modified upper portion of the tubercle at that time appearing as a bract beneath the flower. In Lophophora there is the same condition of things, except that the bract-like upper portion is wanting. From this point of view it would appear that the differences between Lophophora and Echinocactus are intensified by the fact that the flower-bearing areola in the former genus is to be regarded as really lateral on a tubercle the upper part of which has disappeared. This genus occurs abundantly in southeastern Texas, extending southward into Mexico. Mrs. A. B. Nickels reports that the Indians use the plants in manufacturing an intoxicating drink, also for "breaking fevers," and that the tops cut off and dried are called "mescal buttons."

1. Lophophora williamsii (Lem).

Echinocactus williamsii Lem. Allg. Gart. Zeit. xiii. 385 (1845). Anhalonium williamsii Lem. in Forst Handb. Cact. i. 233 (1846).

Hemispherical, from a very thick root, often densely proliferous, transversely lined below by the remains of withered tubercles: ribs usually 8 (in young specimens often 6), very broad, gradually merging above into the distinct nascent tubercles which are crowned with somewhat delicate penicellate tufts, which become rather inconspicuous pulvilli on the ribs: flowers small, whitish to rose: stigmas 4. (Ill. Bot. Mag. t. 4296) Type unknown.

Along the Lower Rio Grande, Texas, and extending southward into San Luis Potosi and southern Mexico.

THE END