carpel, nature of the stylopotlium, winging of tiie fruit, character 

 of the seed-face, number of oil-tubes, and general habit. 

 Thus genera are reached, and in a way that seems to express re- 

 lationship as well as can be done in any lineal arrangement. Oc- 

 casionally genera will be found grouped together which seem 

 farther apart than is consistent with other groupings. This fol- 

 lows from the fact that \rc are only dealing with Nortli American 

 plants, and the intermediate forms not lacing represented in our 

 flora, it hardly seems necessary to multiply subdivisions. It will 

 further be noted that the presentation of genera differs but little 

 from that of Bentham & Hooker, the only radical change being in 

 associating Hydrocotyh- (Tribe I) with Erigoiia (Tribe V), 

 which we have done because of the remarkable lateral flattening 

 of the fruit. 



Directions for collection and study.— It is very evident 

 that no Umbellifer is fit for determination unless it has mature 

 fruit. That this cannot be too strongly insisted upon is shown by 

 the immense amount of immature and hence well-nigh useless 

 material that encumbers our herbaria. The great confusion in 

 which the order has fallen has in no small degree arisen from the 

 attempt to determine the relationships of these immature specimens. 

 No such attempt should be made, for it v;ill be mere guess-work 

 at the best. When mature fruit has been obtained, it is not safe to 

 depend upon surface inspection. This is deceptive even in the 

 matter of oil-tubes, and has led to confusion, and of course does 

 not reach the inner structures which are fully as important. Nor 

 can a section of the dried fruit be depended upon, for the import- 

 ant characters drawn from the seed-face j.re more or less disguised 

 by the drying. The only sure method is as follows: Soften and 

 soak the fruit until it has regained its original plumpness^ and then 

 make a series of thin cross-sections in the middle region of the 

 fruit. Submitting such sections to an inch objective the required 

 characters will be easilv and clearlv seen. 



3 This may be done in a moment or two in boiling water, but it is far better to soak the 

 fruit slowly in cold soft water, with a little potash if it is particularly hard, and in a day or 

 two it will rejjain all its original phunpncss gradually, without any of the violent action 

 that is apt to come from boiling. 



