Swedish Granary/Atlantic White Cedar Project
Cumberland County Historical Society
Greenwich, NJ
Dendrochronology Progress Report
Dr. Edward R. Cook
William J. Callahan, Jr.
November 2014
Introduction
In August, 2008, while engaged on an independent project studying other sites in the vicinity,
dendrochronologists William Callahan and Dr. Edward Cook were asked to assist in the evaluation of a
structure identified as being of historical importance. They therefore collected 10 wood core samples
from a timber building known as the "Swedish Granary", presently located on the grounds of the
Gibbon House at 960 Ye Greate Street, Greenwich, NJ 08328.
In 1973 this structure had been identified by G. Edwin Brumbaugh, PAIA, and Albert F.
Ruthrauff, AIA as being of early Swedish settlement origin, and they suggested a construction date of
circa 1650 and speculated that it might have functioned as a grain storehouse. Also proposing an early
Swedish colonial origin for the building were Carl & Alice Lindborg, who referenced the work of Dr.
Amandus Johnson, author of “The Swedish Settlements on the Delaware, vols. 1&2” (1911). At the
time a decrepit farm outbuilding standing in Lower Hopewell Township, it was disassembled, moved to
and restored at its present site in 1976. It is currently owned and maintained by the Cumberland County
NJ Historical Society (CCHS).
The logs used in the construction are Atlantic White Cedar (Chamaecyparis thyoides), a species
indigenous to a narrow coastal region along the Atlantic, growing in fresh water swamps and bogs.
Although commonly called "Atlantic White Cedar", it is horticulturally not a cedar but a cypress.
However, throughout this report the common name will be employed, often shortened for convenience
to AWC. The wood is lightweight, easily worked and highly resistant to decay, and in southern New
Jersey was used for shingles, barrels, siding, boats, and occasionally as primary timber for construction.
The field sampling of the Swedish Granary resulted in ten wood cores of good quality, but
laboratory analysis afterward was unable to provide dates for the materials. To make precise
dendrochronological datings it is necessary that the field samples be correlated with an absolutely dated
"master chronology" (aka "absolute chronology" or "standard chronology"), which specifically
identifies each calendar year. This master chronology must be compiled previously, of multiple samples
of the same species sharing the same ecological region. In the case of the Swedish Granary, no such
master existed at the time, and therefore no dendrochronological datings were possible.
Although the 2008 sampling did not succeed in providing dates, awareness of the potential of
dendrochronology to contribute irrefutable evidence to the debate about the age of the Swedish Granary
remained. It was realized that the absence of a regional master chronology had precluded successful
analysis, and that a targeted effort could redress the limitation: if a master chronology for AWC could
be compiled, then an absolute dating of the timbers used in the construction of the Swedish Granary
could be possible. To that end, in 2013 the Cumberland County Historical Society NJ applied to and
was granted funding from New Jersey Historical Commission to support the creation of an AWC
master chronology, with the intent to to date the Swedish Granary. Moreover, it was understood that the
project, if successful, would make possible dating of other historical AWC structures in the region.
Theory and Methodology
The word dendrochronology derives from Greek dendron/tree + kronos/a space of time. That
trees develop annual growth rings in response to ecological conditions, and that these growth rings can
be used to align wood specimens chronologically, is an old concept. In the 3 rd century B.C. the Greek
botanist Theophrastus speculated on the correspondence between ecological factors and tree-ring
development, and Leonardo da Vinci noted in the 15 th century that the ring growth patterns of newly
felled trees reflected weather variations that he remembered from his youth. In 1737 French naturalists
Duhamel and Buffon performed a "crossdating" of tree rings based upon a severely frost damaged
growth ring caused by an unusually harsh winter in 1709.
But it wasn't until the 1920's that dendrochronology began to be utilized in a formally scientific
manner. An American astronomer, A.E. Douglass, was searching for correspondence between weather
and sunspots, a solar activity that occurs with more or less cyclical regularity. He hoped to trace the
effects of sunspots in tree rings, which he knew constituted an acceptable record of past climatological
conditions. Although not successful in this experiment, Douglass noted that similar growth patterns
appeared in wood samples collected throughout a very wide area in southwestern USA. In many cases
the patterns of even distant samples could be positioned upon each other like a jigsaw puzzle, some
pieces longer and some shorter, eventually producing an overlapping collection - a "chronology"- that
stretched much farther in time than any one sample.
In theory dendrochronology is simple. Trees expand around their outer diameter as they grow,
and normally each year's expansion becomes visible in the wood as a single ring. Ring development is
a complicated response to local growth conditions, including factors such as temperature, precipitation,
species, soil type, variations in sun, wind, etc.; in general, good ring growth reflects good conditions
and bad growth bad conditions. By measuring the size of the annual rings, i.e. measuring the change in
ring width from year to year, a pattern of growth over time can be established. In theory, all of the trees
in an ecological region share similar growth conditions and consequently all produce similar annual
ring patterns, allowing comparison of their measured patterns with each other. In actual practice,
measurement series from contemporary samples can vary greatly and even duplicate series from the
same tree are never precisely alike. Comparison can prove extremely difficult or even impossible to do.
Dating a sample requires not only a quality measurement but also the previously executed
compilation of local growth patterns over time, the "master" or "standard" chronology. This compiled
measurement represents the mean growth rates for the species and region under study, and is created by
overlapping, more correctly crossdating, measurements from a large number of samples. Generally this
process is begun using samples from living trees, for obviously their outermost rings are of known date.
Specimens are collected from progressively older sources, from buildings or art objects or sub-fossil
logs, etc., which are then measured and mathematically overlapped upon the measurements from the
living trees and from each other. As this process continues a chronology longer than any single sample
is constructed, stretching back in time and absolutely dated by the samples taken from the living trees.
To illustrate simply, assume for example that a researcher in 2010 takes samples from a grove of
200-year old trees. The measurements of its growth rings thus represent the years 1810-2010. Samples
from a nearby structure built in 1910 with 200-year old trees are collected, measured, and added to the
ones from the living tree samples. With a 100-year overlap between them, (1810-1910) a mathematical
mean is calculated from the two sources and the revised master chronology now covers 1710-2010.
Another site is sampled, this one built in 1810 with 200-year old trees, supplying timbers with an
overlap of 100 years on the early portion of the master (1710-1810), and again the measurements are
compiled and a revised mean calculated. Now the master chronology covers 1610-2010. This process is
repeated until there are no samples older than those earliest in the master chronology. In this way some
chronologies, for certain species in certain regions, have been produced that exceed 10,000 years in
length.
Douglass was able to compile a very long master chronology for a species of pine found in
America's southwest, with each calendar year on the the chronology represented by multiple individual
measurements compiled into the standardized mean measurement of growth. He hypothesized that
since each ring in the master represents a single year of growth, any independent pattern that could be
successfully compared to it could be assigned to a specific year by its last growth ring. Douglass used
his master chronology to date to the 13 th century A.D. samples from various abandoned Indian
dwellings in Arizona and New Mexico. After publication of his results, dendrochronology received
widespread attention in the scientific community. Standardized in practice and computerized to assist
with processing the mathematics, modern dendrochronology is practiced worldwide, assisting not only
archeology and architecture, but art history, climatology, geology, hydrology, vulcanology, et alia.
Modern dendrochronology relies heavily on statistical evaluations to aid in synchronizing
crossdating positions between series, and to determine the strength of correspondence between them.
Specialized computer programs evaluate the massive amounts of mathematical data, sometimes
involving thousands of individual series comprising millions of data points, and establish positions of
correspondence exhibiting high degrees of statistical similarity. It is the responsibility of the
dendrochronologist to evaluate this information, for in practice random statistical correspondence
between series may occur at multiple positions, and these "false positives" must be recognized and
disregarded.
The wood samples themselves are collected using a variety of techniques, the particular manner
dictated by the type of object. Most commonly samples are taken using specially designed and
manufactured drill bits, which extract thin cores efficiently and with negligible harm; aesthetic
considerations are satisfied by carefully selecting locations for drilling. Occasionally an artifact can be
taken whole for measurement in the laboratory, or if permissible a piece may be sawn and removed.
Measurement in place can be done using a loupe after some surface preparation, but this method is
extraordinarily time-consuming and inevitably yields imprecise data sets, and so is used very rarely.
In the laboratory the samples are assigned identity numbers and the surfaces carefully prepared
to allow accurate measurement. Each sample is examined ocularly on a movable stage that passes
under a microscope, the operator controlling the speed and distance of the movement. As it passes
under, the researcher uses crosshairs in the reticle to measure each growth ring on the perpendicular, to
a precision of .01mm, and the width is recorded. A single sample may contain hundreds of rings, some
wider than 10mm -a centimeter- and some as small as .02mm - only one or two cells wide. The
measurements are entered into the computer database, easily managed and accessible.
Swedish Granary AWC Project
Status and Results
One inescapable circumstance guided the organization of the AWC Project: without a functional
master chronology, no dendrochronological dating of the Swedish Granary could be possible.
Therefore, development of a suitable standard chronology for Atlantic White Cedar (Chamaecyparis
thyoides) for dating historical structures in southern New Jersey was essential. A complex undertaking,
it would be necessary to locate multiple regional sources of both living and historic AWC, sufficient in
quality and quantity to provide the materials for compiling a viable and representative mean.
To begin, living AWC, preferably in the range of 200+ years old, had to be found and sampled,
and compiled into a viable "live tree" master that would secure the outermost year of the chronology to
2014. Older sources, especially historic structures, had to be located and sampled, and they had to
provide material of such quality and age that the individual "site chronologies" from these historical
buildings (often of undocumented age) could overlap the earlier portions of the "live tree dating
master" with, minimally, a statistically significant 50+ years. Furthermore, the various historic sites had
to be synchronized and overlapped with each other, in order to extend the master in an unbroken
chronological chain sufficiently far back in time to traverse the potential age of the Swedish Granary. In
short, a sequential series of measurements had to be created, affixed in 2014 and continuing
uninterrupted back to approximately 1550, a period covering the purported age of the Granary.
Following extensive planning, field work began in May, 2014. Accompanied by the project's
coordinating manager, Joseph Mathews of CCHS, and assisted by architectural historian Joan Berkey
and historical carpenter J.P. Hand, dendrochronologists William Callahan and Dr. Edward Cook made
more than 20 exploratory and operational sampling visits to the SW New Jersey region. Samples were
collected from multiple stands of living trees in state forests, from sub-fossil trees at various locations,
and from more than 10 historic structures (see Table 1). Nearly a dozen additional structures were
visited and surveyed without being sampled.
The effort to establish the foundational master chronology, one based on living trees and thus
Seq Series Time_span
1780 1800 1820 1840 1860 1880 1900 1920 1940 1960
1839 1859 1879 1899 1919 1939 1959 1979 1999 2019
--- -------- --------- ---- ---- ---- ---- ---- ---- ---- ---- ---- ----
1 NIC01A 1848 2000
.50 .36 .37 .33 .55 .47 .47
2 NIC01B 1844 2000
.26 .38 .38 .47 .51 .56 .53
with each calendar3 NIC01D
year specified,
to the
knowledge,
networks and
1851 2000 succeeded due .46
.42 local
.32 .45
.30 .36 contact
.36
4 NIC02 efforts
1891 2000
.52 .53
.54
thorough reconnoitering
of the personnel from CCHS,
and.55a .52
serendipitous
meeting of
5 NIC03 with
1891
.38 .47Joe.45
.35 .37
colleagues. After contact
the2000
NJ Parks and Forests department,
Mathews
from CCHS located
NIC06
1909 2000
.60 .65 .35 .37
remnant old growth67 AWC
in 1888
Peaslee
Wildlife Management Area.
Permission
was granted, and multiple
NIC07
2000
.47 .54 .72 .70 .69
samples were collected
there
by
Callahan,
Cook,
et
alia.
During
an
informal
conversation
with Nicole
8 NIC12
1903 2000
.55 .55 .59
.60
9
NIC14
1933
2000
.49
.42
.43
Davi, a researcher at Lamont-Doherty Tree Ring Lab, Dr. Cook discovered that she and colleagues had
NIC16 AWC
1902trees
2000 in NJ in 2000, and graciously .22
.24 .30the.31
sampled some old 10
growth
she shared
measurements, some
11 NIC20A2 1820 2000
.47 .53 .51 .37 .31 .18 .35 .37
extending back to121790,
with the present AWC project. Together these efforts provided the necessary
NIC20C3 1790 2000 .12 .24 .35 .43 .38 .44 .34 .36 .36 .37
material for an absolutely
dated,
period
13 CPFF01A
1912live
2013tree master chronology for the .50
.601788-2014.
.60 .41
14 CPFF01B
1891 2013
.09 AWC
.17 .27
.40 It is operationally
Figure 1 shows
the modern
master created by this phase.17of the
project.
15
CPFF03A
1788
2013
.11
.28
.53
.47
.48
.29
.38
.19
.29
.23
robust back to the 1820's, meaning that there are enough individual samples calculated into each year to
16 CPFF03B 1806 2013
.31 .51 .39 .37 .25 .36 .39 .43 .33
produce a statistically
representative mean. Prior
to that decade
the number of sampled trees included
17 CPFF05A 1863 2013
.39 .42 .55 .54 .51 .31
declines, and although
currently
individuals are compiled
of these
18 CPFF05B
1894 several
2013
.54 into
.52 each
.55 .50
.40 earlier years, the
19 CPFF06B
2013 .34
.33 .28 representative
.34 .32 .36 statistically;
.35 .41 .38 before
.32 1800 only three
number is insufficient
to be 1794
considered
rigorously
CPFF07A
1911 2013
.55 longest
.56 .45
trees contribute to20the
chronology.
Nevertheless, this AWC master.61
is the
calendar-year-dated
21 CPFF08A 1913 2013
.56 .53 .45 .39
record of regional 22
annual
growth
for
the
species
Chamaecyparis
thyoides
ever
developed.
CPFF09A 1903 2013
.60 .70 .64 .52
23 CPFF09B 1891 2013
.54 .54 .57 .57 .44
24 CPFF10A 1904 2013
.50 .51 .52 .42
Figure 1
25
CPFF10B
1862
2013
.54
.61
.60 .58 .46 .36
----------------------------------------------------------------------------------------------------------------------------------------------------
AWC Modern Master for Southern New Jersey
In d ic e s
1.5
1
1780
1820
1860
1900
1940
1980
2020
Time in Calendar Years
========================================================================================
Figure 1. Regional Atlantic White Cedar (Chamaecyparis thyoides) dating master for Southern New Jersey covering 17882014. To be serviceable for dating historical structures and objects for the entire European settlement period in the region,
this chronology must be robustly supplemented and extended from its present length to 1550 or before.
=========================================================================
In spite of the important success in establishing this regional live tree master, from the broader
perspective of the AWC project the result remains incomplete. The length in years of the chronology
already demonstrably overlies the normal extent of the natural age expectancy of regional AWC, yet
simultaneously it remains insufficiently long and robust to permit most dating of historical AWC
materials from the early settlement through the post-Revolution era.
As discussed previously, in order to expand the length of master chronologies to years beyond
those of living trees it is necessary to assemble substantial amounts of older materials, identify and
correlate any examples that overlap with the live tree master and/or with each other, and sequentially
build a continuous series of measurements backward in time. Very rarely does this stage produce rapid
progress, since much exertion often leads to no result, and in practice the effort generally is
exceedingly demanding of time, resources, and patience. Requisite materials -the right species, the
appropriate potential age, satisfactory physical condition, etc.- can be difficult or impossible to locate,
and available materials may prove only after much laborious collection and analysis to be qualitatively
or methodologically unsuitable.
As they had in the search for living AWC trees, the associates from CCHS provided efficient aid
in advancing the project. Familiar with local geography and building history, they were able to
compose a list targeting potentially useful structures and objects throughout the region. Over a period
of several months Callahan & Cook visited these sites, surveyed and sampled them whenever the AWC
material was judged promising (see Table 1). In all, several dozen locations were inspected, and
samples of disparate number and quality were acquired from perhaps half of these. Not all the collected
specimens could be utilized, as ultimately many were rejected after examination in the laboratory due
to the poor physical quality of the wood, and/or because they had too few rings to ever reliably support
statistical evaluation: under normal circumstances approximately 60 rings are required to meet minimal
dendrochronological protocols.
Table 1
-----------------------------------------------------------------------------------------------------------------------------------------------------
1) Moses Crossley House, Cape May Court House, NJ
2) Caesar Hoskins House, Mauricetown NJ
3) ۩ Veale-Sewall Cabin, Upper Hopewell, NJ
4) Reeves-Iszard-Godfrey House, vicinity Seaville, NJ (originally Middle Township)
5) ۩ Historical Society Museum Barn, Cape May Court House, NJ
6) ۩ Swing Family Cabin, Mannington, NJ
7) Seaville Quaker Meeting House, Seaville, NJ
8) ۩ Stites Barn, John Gandy Homestead, vicinity Greenfield, NJ
9) ۩ Falkenburge Barn, South Dennis, NJ
10) Schorn Cabin, Swedesboro, NJ
11) Hope House, vicinity Nesco, NJ
12) Richard Townsend House, Middle Township, NJ
13) Penny Watson House, Greenwich, NJ
14) Fred Peech Barn, Marmora, NJ
15) Inn at Historic Cold Spring Village, Lower Township, NJ (originally from Dennisville, NJ)
16) Swedish Granary, Greenwich, NJ (originally Lower Hopewell Township)
========================================================================================
Table 1: Historical sites and structures from throughout south and south-central New Jersey providing AWC wood material
for laboratory analysis. Not all source materials were of sufficient quality to allow subsequent laboratory processing.
The symbol ۩ indicates sites included in the relative master chronology, see Figure 2 and Table 2.
========================================================================================
Fortunately, several selected historical sites provided material of sufficient quality that the
individual timbers could be synchronized with each other and their measurements thereafter combined
into a unified "site chronology". Because they still cannot be assigned specific calendar years they are
designated as "floating" or "relative chronologies", but they remain extremely useful for crossdating
purposes. As mathematical means, such compilations tend to significantly strengthen the statistical
"signal" so that the new combined site chronology more closely reflects actual growth conditions than
does any single individual timber.
Although the selected site series remain undated absolutely, i.e. none of the sites is dated to a
specific year, they all have purported dates suggesting construction in either the 18 th or early 19th
centuries, and therefore were considered as potentially contemporaneous and were tested to discern any
relative overlap. Strong positive correlations between these buildings were discerned and the relative
offsets in the outer, relative dates between buildings were not inconsistent with expectations based on
evaluations of the buildings by architectural historians.
Table 2 shows the statistical strengths of the crossdating of the five buildings purportedly
constructed in the mid-to-late 18th century and the early 19th century. Each site chronology is based on
the average of the measurements of two or more sampled timbers from the site that crossdate amongst
themselves. By first crossdating each building internally, independent of the other structures, a
constraint is imposed on the temporal placements of the different series. The numbers are expressions
of the "r-factor”, the Spearman rank correlation coefficient. This coefficient is a universally standard
measure of relative statistical agreement between two groups of measurements or data. It can range
from +1 (perfect direct agreement) to -1 (perfect opposite agreement). These statistical correlations,
produced by the COFECHA computer program, indicate how well each sample crossdates at various
positions with the mean of the others in the group. The correlations listed vary in statisticalPage
strength,
ZZCOF.OUT
1 of 1
but
all
are
within
a
range
expected
for
correctly
crossdated
timbers
from
the
eastern
United
States.
Printed: 9/28/14, 11:50:24 AM
Printed For: Ed Cook
1
2
3
4
5
6
7
8
9
10
11
12
13
Correlations of
30-year dated segments, lagged
5 years
Table 2
----------------------------------------------------------------------------------------------------------------------------------------------------Seq Series
--- -------1
AWCSFC
2
AWCVSC
3
AWCCMM
4
AWCGSB
5
AWCFKB
--- --------
Time_span
--------997 1077
983 1080
1000 1090
1027 1093
1021 1102
---------
No.
Years
----81
98
91
67
82
-----
with
Master
----0.492
0.497
0.465
0.484
0.362
-----
995
1024
---.46
.51
1000
1029
---.40
.42
.26
1005
1034
---.42
.47
.37
1010
1039
---.41
.52
.39
1015
1044
---.38
.53
.43
1020
1049
---.54
.56
.51
1025
1054
---.56
.65
.69
.53
.24 .38
---- ---- ---- ---- ---- ---- ----
1030
1059
---.55
.67
.70
.55
.51
----
1035
1064
---.71
.62
.66
.76
.53
----
1040
1069
---.71
.54
.66
.61
.52
----
1045
1074
---.57
.47
.64
.55
.46
----
1050
1079
---.55
.38
.59
.54
.43
----
1055 1060 1065
1084 1089 1094
---- ---- ---.33
.51
.49
.42
----
.45
.40
.39
----
.44
.28
.37
----
AWCSFC: Swing Family Cabin; AWCVSC: Veale-Sewall Cabin; AWCCMM: Cape May Museum Barn;
AWCGSB: Stites Barn, John Gandy Homestead; AWCFKB: Falkenburge Barn
========================================================================================
Table 2. Output from COFECHA program showing the statistical strength of crossdating between five independently
developed Atlantic White Cedar (AWC) site masters. The building names corresponding to the series codes are given at the
bottom of the table. Spearman rank correlations between each ‘floating’ master and the mean of the site masters are shown
for 30-year periods with five year, 25-year overlaps. The years assigned are arbitrary (hence ‘floating’) because no calendar
year dates could be assigned by crossdating with a calendar year dated AWC master. However, the offsets of the years of
overlap between buildings are generally consistent with purported construction dates based on historical assessments, circa
1700-1775.
========================================================================================
The black curve in Figure 2 illustrates the composite relative master created from the historical
sites during this phase of the AWC project. The statistical correspondences of the crossdating of the
individual sites in the composite are as listed in Table 2. Key signature years representing uncommonly
narrow ring measurements found in each of the five site chronologies also are indicated in the figure.
Together with Spearman rank correlations, these signature years provide the base alignments for the
crossdating. The dearth of key signature years and the marked visual differences between the
overlapping chronologies are a stark graphical demonstration of the difficulties that have been, and
probably routinely will be, experienced in crossdating AWC historical structures in South Jersey.
Additional evidence of the methodological challenges confounding the AWC project is that only
three of the site chronologies statistically correlate with a t-statistic above the desired t>3.5, the
standard minimum threshold for statistical crossdating. The "t-value" is Student's distribution test for
determining the unique probability distribution for the “r factor”, i.e. the likelihood of the “r factor”
correlation value occurring by chance alone. As a rule, a t=3.5 has a probability of about 1 in 1000, or
0.001, of being invalid. Higher “t” values indicate exponentially increasing, stronger statistical
certitude.
Two correlations exceed >4.0: that is, between AWCSFC and AWCVSC (t=4.9) and between
AWCGSB and AWCFKB (t=4.2); only one other pair of chronologies crossdates with a t>3.5: that is,
between AWCVSC and AWCGSB, (t=3.7). However, when AWCSFC and AWCVSC were combined
into a experimental composite, crossdating between this alternative compilation and AWCCMM and
AWCGSB then exceeded the t>3.5 threshold (t=4.0 and t=3.9, respectively). This exemplifies the
benefit of sequentially building multiple-site, regional master chronologies for crossdating. However, a
project of such thorough breadth requires much effort and time to accomplish, as multiple alternative
configurations must be systematically formulated, tested, rejected or accepted, then supplemented and
adjusted with new alternative configurations, which in turn are tested, and so forth.
Figure 2
---------------------------------------------------------------------------------------------------------------------------AWCCross-Dating Between Five Independent Buildings and
the Regional Average Series w/ Signature Years Indicated
3.5
3
t=4.9
Indices
2.5
2
A
A
A
A
A
A
1.5
1
W C S FC
W C V S C
W C C M M
W C G S B
W C FK B
V E R A G E
t=4.2
0.5
0
980
1000
1020
1040
1060
1080
1100
Relative Time in Years
AWCSFC: Swing Family Cabin; AWCVSC: Veale-Sewall Cabin; AWCCMM: Cape May Museum Barn;
Average Correlation (RBAR) Between the Five Buildings
AWCGSB: Stites Barn, Gandy Homestead; AWCFKB: Falkenburge Barn
[30-year running means w/ 29-year overlaps]
=========================================================================
RBAR
Figure 2. Illustration of crossdating between the five different historical site chronologies described numerically in Table 2.
The building names corresponding to the series codes are given at the bottom of the table. Shared key signature years are
marked with a dotted vertical line. The t-value notation indicates the strength ofRBAR
correlation
=0.42 between specific selected sites.
The series at the bottom -the black curve- is the composite, relatively dated AWC historical master for southern New Jersey,
as presently configured. It is expected to serve as the basis for continued AWC chronology development in the future, with
the intention of eventually connecting the absolutely-dated live tree chronology and the relatively-dated site chronologies
into a continuous master chronology functionally serviceable for dating AWC structures and artifacts over the entirety of the
historical period.
1020
1040
1060
1080
=========================================================================
Relative Time in Years
Remarks
In order to apply dendrochronological dating methods it is necessary to possess a master
chronology for the species and region under study. In the case of the Swedish Granary in Greenwich,
NJ, the lack of such a reference precluded all possibilities of using tree-ring analysis to date the
building when it was initially tested in 2008. The primary structural timbers used in its construction
were of Atlantic White Cedar, a species not compatible with existing oak and pine historical masters. In
the hope of using dendrochronology to date the Granary and other AWC buildings of New Jersey, the
Cumberland County Historical Society initiated a grant funded project in 2013 to create this master.
To construct a statistically robust master chronology is a scientifically complicated and
technically demanding undertaking. Nevertheless, in a remarkably short time and with restricted
resources, unprecedented progress has been made. A modern AWC master chronology, affixed in time
by sampling live trees in the region, now exists for the period 1788-2014. Moreover, a collection of
several "relative" AWC chronologies has been assembled from sampled historic buildings and objects.
Though not yet dated to specific years, their purported dates suggest that collectively they likely can be
assigned to the period circa 1700-1790. At a minimum their existence is additional proof that historical
dating of the species is practicable, and they provide the foundation for future efforts to extend the
master chronology back in time.
Despite this success, the conjunction of the absolute and relative chronologies remains
unachieved. Because of a deficiency of materials from the period bridging these modern and historical
series, no robust composite can yet be compiled. The material available is deficient in both number and
quality. In order to extend the extant AWC masters and thereby establish a continuous chronology
viable over the entire settlement period, it will be necessary to: 1) find more buildings and artifacts
constructed of AWC, cut during 1750-1850, to correlate with the inner portion of the live tree dating
master, and which also offer enough rings to bridge the gap between the dated master and various
undated site masters; 2) find in the landscape remnant AWC logs from trees that died decades ago, socalled sub-fossil material, and sample these for eventual addition to the dating master.
No buildings from the late 18th through mid 19th centuries containing AWC of suitable quality
have been identified and sampled, but it is hoped that with time, exploration, and advertising some may
be found. Experience hereto has been that the latest structures to contain sizable AWC timbers were
constructed by the first decade of the 1800's. As a consequence, living trees with 250+ rings would
have had to be available in order to have sufficient overlap with the ring series from the historic
structures. None so old have been found, apparently because intensive harvesting of AWC for
construction during the colonial era had exhausted supplies of old-growth AWC in the region. Of
course, old trees may still exist in the forests, but a search will require much more exploration than the
timetable and funding of this project allowed.
Two large AWC logs lying on the surface of a coastal salt marsh were sampled. Each of these
logs had over 200 rings but neither crossdated with the dating master shown in Fig. 1, nor did they
crossdate with any of the collected historical samples. Remarkably, the logs did not crossdate with each
other either. Although in a similar location and visibly in similar states of physical preservation, their
ages in fact may be very different from one other, perhaps by hundreds of years. Logs of this kind can
serve as a valuable source of rings and should be sought, for if enough are found some will contribute
to extending the master chronologies, potentially back centuries into pre-settlement times.
Although much has been accomplished, one goal of the AWC project remains unfulfilled:
successfully dating the Swedish Granary. Repeated attempts to correlate the Granary with the modern
chronology and/or site chronologies have procured no viable result. Reasons for this are purely
speculative at this time. It is possible that the age of the building in fact is as proposed by some
architectural historians, i.e. in the later part of 17th century, in which case it lies outside the present
range of existing chronologies. Another potential explanation is that it is similar in age to tested
historical sites, i.e. post-1700, but that the signal of its internal site chronology is statistically too weak
or anomalous to correlate at present with any existing masters; the compiled mean chronology for the
Granary timbers contains in toto just 69 rings, a not overly large number if synchronizing with other
non-robust tree ring series. Alternately, the Granary could be from the 18 th century or even the 19th
century, and the same conditions of statistical weakness could preclude crossdating under current
circumstances.
It seems very likely that the Swedish Granary ultimately can be dated successfully using
dendrochronology. Experience from the recently completed stages of the investigation proves that
AWC can crossdate and that master and site specific chronologies can be compiled thereafter. However,
creating a wholly new operative dating chronology for a previously untested species requires resources
and patience. It took A.E. Douglass seven years to finalize the master chronology that enabled him to
date the Anasazi dwellings in the American Southwest. The historical master chronology for oak
compiled by Cook & Callahan had its beginnings in eastern forests in 1984 and was not fully functional
until 1992. In comparison, since the active field sampling phase of the AWC project began in summer
2014 it has proceeded with extraordinary speed and efficaciousness. The results are very promising.
Edward Cook was born in Trenton, New Jersey, in 1948. He received his PhD. from the Tucson Tree-Ring Laboratory of
the University of Arizona in 1985, and has worked as a dendrochronologist since 1973. Currently director of the Tree-Ring
Laboratory at the Lamont-Doherty Earth Observatory of Columbia University, he has comprehensive expertise in designing
and programming statistical systems for tree-ring studies, and is the author of many works dealing with the various
scientific applications of the dendrochronological method.
William Callahan was born in West Chester, Pennsylvania, in 1952. After completing his military service he moved to
Europe, receiving his MA from the University of Stockholm in 1979. He began working as a dendrochronologist in Sweden
in 1980 at the Wood Anatomy Laboratory at the University of Lund, and returned to the United States in 1998. A former
associate of Dr. Edward Cook at the Tree-Ring Laboratory of Lamont-Doherty, he has extensive experience in using
dendrochronology in dating archaeological artifacts and historic sites and structures.