396 FILICALES [CH. 



The difficulty is that the evidence of reproductive organs is 

 very far from decisive. In the absence of the female repro- 

 ductive organs, the seeds, we cannot in most cases be certain 

 whether the small sporangium-like bodies on fertile pinnules 

 are true fern sporangia or the microsporangia of a heterosporous 

 pteridosperm. What is usually called an exannulate fern 

 sporangium, such as we have in Angiopteris and in many 

 Palaeozoic plants, has no distinguishing features which can 

 be used as a decisive test. The microsporophylls of the 

 Mesozoic Bennettitales produced their spores in sporangial 

 compartments grouped in synangia like those of recent Ma- 

 rattiaceae; and in the case of Crossotheca, a type of frond 

 always regarded as Marattiaceous until Kidston' proved it to 

 be the microsporophyll of Lyginodendron, we have a striking 

 instance of the futility of making dogmatic assertions as to 

 the filicinean nature of what look like true fern sporangia. In 

 all probability Dr Kidston's surmise that the supposed fern 

 sporangia known as Dactylotheca, Renaultia, Urnatopteris are 

 the microsporangia of Pteridosperms will be proved correct^. 

 The question is how many of the supposed Marattiaceous spo- 

 rangia must be assigned to Pteridosperms ? There is, however, 

 no reasonable doubt that true Marattiaceae formed a part of 

 the Upper Carboniferous flora. All that can be attempted in 

 the following pages is to describe briefly some of the numerous 

 tj^es of sporangia recognised on Palaeozoic fern-like foliage, 

 leaving to the future the task of deciding how many of them 

 can be accepted as those of ferns. It is impossible to avoid 

 overlapping and some repetition in the sections dealing with 

 true Ferns and with Pteridosperms. The filicinean nature of 

 the stem known as Psaronius (see page 415) has not as yet 

 been questioned. 



The nomenclature of supposed Marattiaceous species from 

 Carboniferous and Permian rocks is in a state of some confusion 

 owing to a lack of satisfactory distinguishing features between 

 certain types to which different generic names have been 

 assigned. As we have already seen in the case of supposed 



1 Kidston (06). = Kidston (06) p. 429. 



