PBEFACE. IX 



families or genera within the limits of a fauna may, when all the forms 

 of an entire continent or zoogeog-raphic '"region."' or the world at large, 

 are examined, be found to be connected by intermediate " extralim- 

 ital" forms. Sometimes, however, this test proves exactly the reverse 

 to be true. Therefore, in the present work the families and genera 

 recognized have not been based on the species belonging to North and 

 ]VIiddle America alone, l)ut on all others that were available, so far as 

 time permitted. 



It is often difficult to weigh accurately the value of structural dif- 

 ferences; there are many cases in which the author has long remained 

 undecided what course to adopt, but decision, one way or another, has 

 been necessary, and it only remains to be said that in such cases the 

 l)enetit of any doubt has been given to established usage, in order not 

 to disturb current nomenclature by unnecessar}^ innovations. 



The question of whether a given form should be considered as a 

 species or a subspecies is very much a matter of material, botli from 

 a geographic and a numerical point of view. The greater the num])er 

 of closely related forms, hitherto regarded as specifically distinct, that 

 are examined — especially when representing intermediate localities — 

 the fewer becomes the number of those which are really specilically 

 distinct. As in the case of genera, very different extremes are often 

 connected by a series of intergrading forms, approaching one or 

 the other of the extreme types exactly in proportion to their geo- 

 graphic position between them; and other forms much less different 

 appear to be really distinct through absence of "intergrades."" In 

 di^termining questions of this class the author has exercised the fullest 

 independence, without reference, so far as North American forms are 

 concerned, to the rulings of the committee^ of the American Ornithol- 

 ogists' Union; ^ not from lacic of confidence in the committee's judg- 

 ment, })ut from a full knowledge of the unsatisfactory conditions as 

 to time and material under which their conclusions were usually 

 reached. Satisfactory decisions affecting the status of described but 

 still dubious forms is a question both of material and investigation, 

 and the author holds that no conclusion in such a matter should be 

 accepted unless based upon an amount of material and careful investi 

 gation equal to that bestowed by the original describer. 



Recognizmg the fact that in the present stage of zoological nomen- 

 clature trinomials are a ''necessary '?vil,'' the author has not hesitated 

 to use them when such relationship was clearly indicated by the evi- 

 dence. He has not, however, often done so on theoretical grounds, 

 ])ecause, in the first place, the facts when known may or ma}^ not 

 justify the step, and in the second because a binomial is preferable to 

 a trinomial when there is anj^ good excuse for its adoption. The 



^ As set forth in the Check List of North American Birds (editions of 1886, 1889, 

 and 1895) and various supplements to the same. 



