RINGS 33 



ring is greatly expanded in one direction and somewhat diminished 

 on the opposite side. This gives the appearance of a series of cres- 

 cents on one side of the tree section. It usually interferes completely 

 with the rain record in the tree, but at the same time has a strong 

 climatic significance as an indicator of heavy spring winds. 



Other false rings — Other abnormal rings are sometimes produced. 

 Sequoia radials occasionally show certain "pitch" or "pith" rings. 

 These are white, very narrow, and totally different in color from the 

 rest of the wood. If they seem very soft, they have been noted as 

 pith rings; if hard, as pitch rings. They may come either within a 

 year's growth or between two years. They therefore are very annoy- 

 ing, for they destroy the count, it being impossible to tell whether the 

 normal rings on each side belong to one year or to two. I have made 

 it a rule to discard entirely regions of ring record thrown into doubt 

 by such rings. Doubtless they come from injury and usually from 

 fires. In the yellow pines no similar rings have been noted, but in 

 each tree abnormally large rings occur close to large fire injuries during 

 the early period of recovery and diminished rings in other parts of 

 the tree circuit. 



Effect on means — In all cases of ring errors that leave any uncer- 

 tainty in dating, the uncertain part, or even the whole tree, is omitted 

 from the means. In large groups, of course, the omission of a tree is 

 usually a small matter, but in the early years of the group record it 

 may be serious, for the number of individuals decreases as we go back 

 to earlier and earlier dates. In such cases only the uncertain part 

 is omitted. But here another difficulty is introduced, namely, the 

 break in the averages at the beginning and end of the omitted part. If 

 the tree in question agrees very closely with the mean of the rest in 

 size of rings, the break does not introduce error; but if it is very 

 different, it has to be merged with the average of the rest in some way. 

 This becomes the same problem as that of introducing a tree of late 

 starting-date into a long group record. 



