THE ANGIOSPERMAE 1371 



intermediate forms are best described by means of the accompanying 

 diagram, after Erdtman (Fig. 1273). 



A special vocabulary of descriptive terms for various parts of the pollen 

 grains has been devised, by means of which the grains can be described with 

 something like the precision of a specific diagnosis. It is too detailed to be 

 reproduced here, but those interested can find the terms defined in Erdt- 

 man's " Introduction to Pollen Analysis " or Faegri and Iversen's " Modern 

 Pollen Analysis ". The description of grains with the necessary exactitude 

 presents certain difficulties, for not only must the grain be viewed from 

 several different angles in relation to its polar axis, but the appearance of the 

 grains, whether in the fresh state or after they have been allowed to swell in 

 contact with a watery mounting medium, is very different. 



The sizes of pollen grains are as variable as their forms. Erdtman groups 

 them as follows: 



Perminuta <iO[x Magna 5o-ioo[j. 



Minuta io-25[ji, Permagna ioo-200[j. 



Media 25-5C!jl Gigantea >200ia 



Extreme sizes are rare. Among familiar plants the smallest pollens are 

 those of Myosotis alpestris (2-5-3 -5^.), Echiiim viilgare {io-\^\x) and Urtica 

 urens (i4[x). At the other end of the scale are large grains such as those of 

 Ciiciirbita pepo (230[j.), Mirabilis jalapa (250a) and Elodea (134^1). The 

 largest grains are associated with a relatively small number in each pollen 

 loculus. Mirabilis has only 32 grains per loculus, whereas in Borago offici- 

 nalis, which has grains measuring only about S'^, the number per loculus 

 may be 60,000. The largest grains seem, as a rule, to be produced by 

 ephemeral flowers which last only for a single day. 



The systematic importance of pollen form should not be overrated 

 despite the variety and intricacy of the grain structure. A wide classification 

 based on pollen types would be artificial, but within limited circles of affinity 

 the evidence they afford may often be very valuable. Most genera show a 

 marked consistency of pollen form, though there are some striking excep- 

 tions. The generic identification of pollen is usually possible, but specific 

 identification, with any reasonable certainty, is much less generally feasible. 

 Exceptionally, we find pollen of two or more types even within a genus, 

 but this has not been hitherto accepted as a sufficient ground for subdivi- 

 sion of the genera concerned. A good example is Tiilipa, a genus with very 

 constant floral characteristics, almost certainly a natural genus, but having 

 monocolpate pollen in some species and tricolpate pollen in others. It is 

 the only Monocotyledon with this latter type of grain. Again the genus 

 Crocus, also a very natural group, has pollen either without any colpae or 

 pori, or else with several, parallel, ring-like furrows. Polygonum grains have 

 either a single median porus in each colpa or pori free on the surface, not in 

 colpae. 



In some families there are characteristic peculiarities which may be 

 sufficiently constant for use in diagnosis. Thus Cucurbitaceae have all very 



