TREE-RING INDICES OF RAINFALL, TEMPERATURE, AND RIVER FLOW 1025 
Methods [8]. Some special techniques of dendro- 
chronology which have been found of value but which 
can only be mentioned here, are (1) use of the Swedish 
increment borer, which permits wide and rapid core 
sampling, (2) mounting and surfacing of cores in such 
a fashion as to permit ease and rapidity of cross-com- 
parison of gross features as well as good definition of 
cell structure under the microscope, (3) elimination of 
individual age trends (varying mean growth rates) in 
order to average the ring records in uneven-aged trees 
of a set and thus remove a part of the ever-present 
random term in the growth of each tree and make the 
data for different time intervals comparable, (4) use of 
extensive areal averages to remove still more of the 
random term in the mean growth at each site, and (5) 
use of skeleton plots as a reconnaissance tool in archaeo- 
logical dating. 
Errors in the Unit. Possible errors in dendrochrono- 
logic series are of two types: (1) identification of growth 
irregularities, such as locally absent or false rmgs, and 
cypress, are so subject to false-ring layering as to be 
unusable in chronology building. 
Errors of Interpretation. It has long been accepted 
as a botanical axiom that the absolute growth of plants 
is subject to a matrix of environmental and transmitted 
factors, whose imteractions are extremely complex [15]. 
Even when the problem is simplified by the use of 
growth departures, it appears that only on sites critically 
limited with respect to an important growth element, 
such as rainfall, can such departures be highly repre- 
sentative of that element. There may be hidden effects 
in the growth record of past fires, pest outbreaks, 
mechanical injury, changes in associated flora or micro- 
organisms, and other factors. Some species are char- 
acterized by an inherent tendency to nonuniform cam- 
bial activity, so that the ring series along any radius 
seems to have no rational explanation [24]. 
Some meteorological and hydrological factors, too, 
tend to destroy any simple relation between seasonal 
rainfall and tree growth: variation in distribution of 
370 600° F ap 
MEAN TREE GROWTH 
FLAGSTAFF 
1stG) 
MUSA VEROE 
0” 6b 7 Be 730) 760 
AJ] MESA VERDE 
Fic. 1.—Comparison of the mean departures in tree-ring width for two localities in northern Arizona and Mesa Verde 
in southwestern Colorado. 
ical beams. 
(From Tree-Ring Bulletin.) 
(2) interpretation of the ring fluctuations in terms of 
climatic variation. 
Tn particularly severe years much of the cambial area 
may be entirely inactive in terms of new cell growth. 
Locally absent rings may occur with a frequency of as 
much as 5 per cent or more in the most sensitive and 
slow-growing ring series in semiarid regions [27]; they 
are of rare occurrence in more humid regions [17], at 
upper timberline, and in the Arctic. Such rings are 
identifiable by comparison of approximately parallel 
patterns m climatically related trees of a site or area. 
False annual rings are comparatively rare in conifers 
in high latitudes and near upper timberline, but are very 
common in the latitudes of southern Arizona. Non- 
traumatic false rmgs are found in general to occur as 
a function of species, climate, environment, and age. 
In such dendrochronologic species as Douglas fir and 
ponderosa pine they may be uniquely identified by 
crossdating as well as (in almost all imstances) by 
absolute criteria [25]. Some species, such as Arizona 
The records in living trees were extended backwards by use of overlapping series in archaelog- 
Smoothed curves were superposed to permit greater ease in visual comparison of the longer fluctuations. 
rainfall within the season significant for ring growth, 
varying influence of temperature and other meteoro- 
logical elements, real differences in total rainfall between 
the tree sites and the correlated meteorological station, 
and the numerous influences affecting the rainfall-runoff 
relationships from basin to basin. 
It is thus a fundamental requirement for the pro- 
duction of significant dendroclimatic series that the 
ring record of any tree be checked against sufficient 
associated trees, and that the group mean, in which 
random errors have been permitted to cancel as far as 
possible, in turn be checked against and averaged with 
a sufficient number of associated groups. 
Relation to Neighboring Sciences. Like most work 
outside the classical disciplines, dendrochronologic re- 
search is closely related to a number of fields: botany, 
ecology, forestry, meteorology, hydrology, archae- 
ology, geology, and astronomy. The analyzed vari- 
able, tree-ring width, is in the domain of the first 
three of these fields. Its application to archaeology has 
