204 TABLES AND RESULTS OP THE PHE C IPI T A TION, 



Diagram 3. 



The lines of dashes connecting the several type-cnrves on the preceding diagram (2) 

 are an attempt at further generalization by showing the existence and connection 

 of a principal maximum in summer, followed by a minimum in autumn. We 

 can further trace out a secondary maximum late in the year, and a minimum, 

 sometimes of a primary, sometimes of a secondary character, about the second 

 month. 



Large tracts of country are still destitute of stations of long series, but covered 

 with a sufficient number of stations of short series to enable us to make out the 

 character of the annual fluctuation. The following stations have been thus com- 

 bined into groups : — 

 Group 1. — Central Pennsylvania. 10 stations: Bedford, Somerset, Fleming, Moss 



Grove, Butler, Huntingdon, Shamokin, Harrisburg, Carlisle Barracks, 



Canonsburg. 

 Group 2. — West Virginia and adjacent part of Virginia. 7 stations: Winchester, 



White Sulphur Springs, Ashland, Lewisburg, SheetsmiUs, Alleghany 



Coimty, Anthony's Creek. 

 Group 3. — Southeastern Virginia. 5 stations : Williamsburg, Crichton's store, 



Genito ^lills, Smithfield, Powhatan Hill. 

 Group 4. — Western Florida. 4 stations : Fort Barrancas, Warrington Navy Yard, 



Pensacola, Barrancas Barracks. 

 Group 5. — Alabama and adjacent part of Mississippi. 4 stations: Huntsville, 



Greene Springs, Greensboro', Columbus. 

 Group 6. — Central Texas. 5 stations: Fort Belknap, Fort Chadbourn, Fort 



Mason, Fort McKavett, Fort Lancaster. 



