Crossdatingin Dendrochronology A. E. Douglass Laboratoryo! Tree-RingResearch, University o! •4rizona This paper definesand illustratescrossdating, an initial processin dendrochronologyor tree-ring work by which accuratering chronologiesmay be built for dating purposes,for climatic information, or for certain ecologicalproblems. Here are briefly explainedits operationby an efficientmethod, its principles of interpretation and application, its character as differentiated from correlation, its proceduresfor reaching assurancein results, its significanceas a guide to special sites where certain climatic effects on tree rings can be distinguished, and finally references are given to some of its published discussions.The purpose of this paper is to call the attention of ecologists and others to this fruitful processthat carries convictionby tests on well-located trees but whose reality in certain well-assuredregions cannot be judged by misinterpretation of material or un- technical treatment of specimens. HEarti,c, lecalled "The Dendrochronology wcre the only onesthat had what we consider sensitivity,even though to a very small degree; berEnigma 1940 JOURNALbyA. W. OF Sampson FORESTRYintheis basedDecem- on and from that peculiarityhe becameuncertain of a seriousmisunderstanding of the principleon ring identity obtainedby countingin from the which dendrochronologyor tree-ring work is outside and omitted them. Those Davis did use founded. Our criterion of climatic character in had practicallyno sensitivityand, therefore,had the ring recordsin our trees is not sensitivity no patternsto crossdate,and thusno evidentcli- (the amountof fluctuationin thicknessfrom ring matic value so that negativeresults were to be to ring) as he statesand criticises,but crossdat- expected. ing qualitywhich is of an altogetherdifferent na- In localitiesnew to us our standardprocedure ture, Crossdatingis the recognitionof the same is to securea seriesof ring sequencesfrom incre- ring pattern in differenttrees, so that the actual mentcores or stump-cutstaken out in the fieldby growthdate of any onering of the patternis the someonewho understandsthe effectsof topogra- samein the differenttrees and onemay carry a phy, in order to see what climatic effectsthe chronologyacross from tree to tree. Sensitivity trees of the region are showing normally is a seconklaryfeature which is importantonly whengrowing in an environmentcapable of pro- in the presenceof crossdating.If thereis no sen- ducingclimatic effects in the rings. In the South- sitivitythere can be no patternsand thereforeno westwe seeksteep slopes and shallowsoils and a crossdatingby patterns,and no assurancethat rainfall about as small as the trees can stand. the specimencame from a sitein whichthe trees These are of course not the characters a lumber- canshow a climaticvariation year by year. manprefers; in fact he avoidsthem. In the study Inattentionto the initial crossdatingby pat- of large numbersof suchspecimens, we had be- ternswas the error in the studyof 21 increment comefamiliar with the normal Flagstaffpatterns coresby W. E. Davis• someyears ago, cores that and couldrecognize at sightthat eventhe better camefrom the SouthwesternForest and Range specimensdiscarded by Daviswere still not good ExperimentStation at Flagstaff,Arizona, and enoughto be usedin his test. Theseerrors would which ! later examinedvery carefullytogether have been avoidedby his adherenceto the re- with his reportmade available through that sta- quirementof initial crossdatingin real ring pat- tion. He discardedabout half of thosespecimens terns. becausehe was not sure of the identityof the It is surprisingto mostpeople to seecrossdat- rings. While this was apparentlya worthy rea- ing in the form in which we can show it. Our son,the fact wasthat the specimenshe discarded success is believed to be due to favorable sites in a dry region. Advantageousconditions reveal •Co-authorwith A. W. Sampsonof "Experimentin themselvesin certain superblocations in Mesa Correlation of Tree Growth Rings and Precipitation Cycles,"published in Trans. Amer. Geoph.Union, 493- Verde and near Durango,Colorado and many 496, 1936. This paper is referred to in "The Dendroch- otherplaces. Figure ! showsa very distinctive tonology Enigma." Mr. Edmund Schulman of this lab- topographicenvironment in the Mesa Verde Na- oratory in March 1939 examined substantially all of the collections and analysis on which that article was tional Park, one of the best sitesseen in forty based and found tundamental errors that showed the years of field work. The available moisturefor necessityof a completerevision of that paper. In the text above I am giving my personal experiencewith an the fir treeson the slopeis reshic?edto the pre- earlier Davis test. cipitation that falls on or near the trees them- 825 826 JOURNAL OF FORESTRY Fig. 1. Mesa herde site in Fewkes' Canyon. highly favorable to crossdating;the mesa lop at the right slopes generally away from this canyon. CROSSDATING IN DENDROCHRONOLOGY 827 selves,chiefly in the winter rainy season;the "Hopi" signatureas it is commonin the beams steepslope reduces local conservation of moisture of the Hopi villages.One can start here with the to a minimum; an elevation of 6,700 feet above two large rings in the left centerof each,1530- the sea and a mean annualrainfall of perhaps 1531,and especially the smallring of 1532; then twenty incheskeep thesetrees under a stressof list the smallrings both backwardand forward moisturedeficiency. Our incrementborings in and observethe agreementsbetween these trees. five of thesetrees, well distributedboth as to lo- That is the sort of crossdatingwe havefound in cationand age of tree, showa very high quality thousands of cases. of crossdatingin the ring patternsand a sensi- tivity that we havecalled class AA. But most people are familiar with plotted Peoplewho live in moist climateshave great curvesand so Figures 4 and 5 have been ar- difficukyin judgingthese topographic characters rangedto giveplots of measuresof the thickness as such sites are rare and experienceis often of thesesame rings shownin Figures2 and 3. neededin evaluatingthem. The chief requisites The similaritybetween these curves is evident.In hereare: first,the near impossibilityof the trees a regionwhere moisture is the criticalfactor, the getting"imported" water that has fallen on, or years of minimum growth seemsto be more in- been stored in, other localities and then moved dividualizedand to afford more ready pointsof to the area whose trees are under test; second, a correspondencebetween different trees. To these shallowsoil or one with small water storageca- examplesof crossdatingin plotsis addedFigure pacity; and third, a low mean precipitationso 6 showingthe agreementbetween two groupsof that the treesrarely or neverget all the moisture treesabout 70 milesapart. The uppershows the they can use. averagegrowth in seventrees in the Chuska Mountains between Fort Defiance and C,hinle. ILLUSTRATIONS OF CROSSDATING The lower curve is from a group of three logs Our crossdatingis done directly in the ring that camefrom near the storeat Pinyon in the patternson the wood (often aided by what we middle of the Black Mesa, about 70 miles west of call "skeleton"plots), hencein spiteof their un- the othergroup. familiarity direct photographsof ring sequences Finally Figure 7 givesa generalcurve from 55 constituteour first exhibit. Figure 2 represents treesin the northernparts of Arizonacompared patternsin four differenttrees of a sequenceof to the winter precipitationrecords from three yearsin the middle 1200's. We have calledthis widelyseparated points in the samearea that are patternthe "Gap" signaturebecause these rings highlyrepresentative of the area. The winterrec- were involved in the successful1929 dating ord is heremodified by introducinga lag of 2•/• acrossthe gap betweenthe earliesthistoric dates yearsin the smoothedvalues, which means that andthe mostrecent prehistoric ones. In examin- annualchanges are retainedbut that a conserva- ing this illustrationone shouldstart with the tion is introduced in the smoothed values. This threelarge rings near the left end of eachphoto, is a slight variant of the author'susage in test- the rings dated 1248, 1249, and 1250, and note ing Prescotttrees, published in 1919.a in eachtree the smallrings at 1251, 1254, 1258, 1263, 1270, and 1276 (this last in two only). The CROSSDATINGMEMORY illustrationwould be more appealingif rings and The most efficient and at the same time the dateswere not markedand onehad to "dig" out mostconvincing method of crossdatingis by that this similarityfor himself. Thus he wouldtake memorywhich develops on examiningscores of the smallring 1251 and then make a list of the specimensof approximatelythe same age in datesof succeedingvery smallrings for eachtree whichsimilar patternsare identifiedin the great andfind the agreementbetween the trees. majority of cases.Usually the processbegins in Figure 3 representssequences from four trees the selection of some feature such as a little com- that grew in the 1500's.2 We havecalled this the binationof ringsthat is foundto be commonto aFigures2 and 3 are photographsof specimenscol- two or three specimens;then some number, a lected under the auspicesof the National Geographic real dateor a hypotheticaldate is appliedto one Society. My sincere thanks go to that society, to the Carnegie Institution of Washington, to the University of Arizona and other institutions and personswithout SSeealso F. P. Keen, "Climatic