HANDBOOK OF THE BROMELIACE^i. 27 



appeared as a distinct genusiu Mr. Baker's revision of 1888. Other- 

 wise, the division with its five genera is arranged as in the ' Genera ' 

 Plantarum.' Wittmack's Tillandsiea has eight genera, for he re- 

 cognises as distinct Massangea, also a subgenus of Caraguata in the 

 'Hand-book/ and Schlumbergeria, considering the variations in 

 length and coherence of the perianth and staminal whorls to be of 

 generic value. He also separates Vriesea from Tillandsia. 



Having gone steadily through the book, while arranging by it 

 the BromeliacecB in the Herbarium of the British Museum, we may 

 perhaps be allowed to point out a few matters which suggest 

 criticism. 



The key to the Genera with which the book opens, and those 

 to the species with which all the larger genera are supplied, are of 

 great assistance in naming specimens, but now and again there 

 comes a little hitch in the arrangement. For instance, in the key 

 on p. x., the variations in relative length of sepals and petals are 

 used as distinguishing characters between the four genera, Strepto- 

 calyx, JEchmea, Billbergia and Quesnelia, and those of sEchmea are 

 said to be " usually not much longer than the sepals." But on 

 p. 32, in the diagnosis of the genus, we read, " petals two or three 

 times the length of the sepals " ! A glance through the descrip- 

 tions of the species shows that the relation between the lengths is 

 very variable, the petals may be only " shortly protruded" in one 

 species, and " twice as long as the sepals " in the next. Again, in 

 the key to the Subgenera of the group Tillandsiece, Diaphoranthema 

 heads the list of those with rosulate leaves, but the leaves, though 

 often tightly packed on the stem, apparently never form a rosette. 



There seems to be a curious fatality about Mandon's numbers 

 which are quoted, the plants often varying considerably in detail 

 from the printed description. Thus Mandon 1174 is Tillandsia 

 biflora Kuiz & Pav., and ought therefore to have the panicle 3-4 in. 

 long, the lower branch-bracts 1^- in., and the flower-bracts % in. 

 long ; but in Mandon's specimen at the British Museum these are 

 respectively 8 in., 5-6 in., and § in. Similarly, 1177, T. capillaris, 

 has leaves nearly 2-| in. instead of 1 in. In T. virescens, on the 

 contrary, the leaves and peduncle only reach half the lower 

 measurement given. Small discrepancies of this nature are not 

 infrequent, and of course must be expected to a certain extent, as 

 plants will vary, but this marked association with one particular 

 set is curious. Pitcaimia asterotricha Popp. & Endh, and Puga 

 grandiflora Hook., are regarded as synonyms of Pitcaimia ferru- 

 ginea Buiz & Pa von, but the petals of the species are' described 

 merely as yellowish white ; this is true for P. grandiflora only, 

 Popp, & Endl. have "dilute livide purpureis," and Buiz & Pavon 

 " sub-purpureis." 



In conclusion, we must thank Mr. Baker for so frequently 

 quoting numbers, the full value of which is best appreciated when 

 they are wanted ; but to quote all would add considerably to the size 

 of the book, and this might not be desirable. 



A. B. Bendle. 



