cont. from page 1
In total, 17 samples were collected for
tree-ring dating of the DuBois Fort. In every case, the numbers assigned to
the oak samples were the same as those inscribed on the logs for historical
documentation. See Figure 1 and
Table 1 for details.
The wood core samples were processed following
well-established methods of dendrochronology. They were taken to our Tree-Ring
Lab where they were carefully glued onto grooved mounting sticks. The wood cores
were than sanded to a high polish to reveal the annual tree rings clearly. The
rings were than measured to a precision of ±0.001 mm.
The actual cross-dating
procedure involved the use of a computer program called COFECHA (Holmes 1983),
which uses a sliding correlation method to identify probable cross-dates between
tree-ring series. Experience has shown that this method of cross-dating is
superior to that based on the skeleton plot method (Stokes and Smiley 1968) for
oaks growing in the northeastern United States. It is also very similar to the
highly successful CROS program used by Irish dendrochronologists to cross-date
European oak tree-ring series (Baillie 1982),
We used COFECHA to first establish internal or
relative cross-dating among the house timbers. This step is critically important
because it locks in the relative positions of the timbers with each other and
indicates whether or not the dates of those specimens with outer bark rings are
consistent. Having done this, we compared the internally cross-dated series with
independently established tree-ring chronologies from old living trees and
historical tree-ring material. All of the “dating masters” used are completely
independent of the samples taken from the DuBois Fort.
Results
The results of the dendrochronological dating of
the DuBois Fort oak timbers are summarized in Figure 1-Figure
2 and Figure 3, with details
on the dating of each timber given in
Table 1.
A. Middle and Rear Cellar Rooms
Of the eight oak joists sampled for dating, six
provided cutting dates of 1703 (two in the rear room and four in the middle
room; see Figure 1). This result Is extremely robust (see
Figure 2). The
mean chronology of these timbers has a correlation of 0.75 with the dating
master developed from the Jean and Abraham Hasbrouck houses. Indeed, the two are
so similar as to make it a virtual certainty that the oaks used for the
construction of all three houses came from the same local forest or woodlot. The
1703 date also means that these two cellar rooms were part of the original
construction of the DuBois Fort. In all likelihood, the oaks were cut in late
1703 or early 1704, with construction completed sometime in 1705 per the
historical documentation.
Two joists in the rear cellar room did not
cross-date with the dated joists, nor did they cross-date with any oak dating
master in the Hudson Valley, northern New Jersey, and eastern Pennsylvania. The
reason for this is unclear. In addition, none of the oak logs lying on top of
the fireplace hearth cross-dated with any master. These logs may have come from
a totally different location and at a later date.
B. Front Cellar Room
Of the five oak joists sampled for dating, three
provided firm cutting dates of 1835, while one produced a cutting date of 1836
(see Figure 1). The fifth joist did not date out. The fireplace lintel
produced an outer-ring date of 1832. This slightly earlier date may reflect the
loss of some sapwood rings during the creation of the lintel.
Compared to the middle and rear cellar rooms, the mean chronology of the front
cellar room cross-dated much more weakly with
oak dating masters from New Paltz (the same used in Figure 2),
northern
New Jersey, and eastern Pennsylvania. Yet, all produced the same outer-most date
of 1836 with a statistical significance p<0.0O1.
The best result, using the eastern
Pennsylvania master, is shown in Figure 3.
With a correlation r=0.47,
the result visibly weaker than that shown in Figure 2, but still highly
significant statistically. By the time of the construction of the new wing
to the DuBois Fort, it is likely that the local supply of building timber
was depleted. In addition, local commercial sawmills were probably producing
dimensioned lumber for construction from a wider variety of timber sources.
Consequently, the weaker cross-dating for the front cellar room is not’ that
surprising. Yet it is still firm enough to say with certainty that the
DuBois Fort addition was constructed sometime after 1836.
Conclusions
Based on a dendrochronological
analysis, the original portion of the DuBois Fort was constructed from trees
cut in late 1703 - early 1704. This result strongly supports the 1705
construction date that was obtained from numerals on the iron rods used to
reinforce the stone walls of the house. The tree-ring chronology obtained
from the DuBois Fort oak timbers agrees remarkably well with those from the
Jean and Abraham Hasbrouck houses. This indicates that all three houses were
built using the same oak timber source.
The addition to the DuBois Fort over the front cellar room
was constructed after 1836. The quality of the cross-dating is much weaker,
but still unequivocal. The fact that the tree-ring chronology from this room
cross-dates equally well with a number of master chronologies in the Hudson
Valley, northern New Jersey, and eastern Pennsylvania suggests that the wood
used for construction was probably not local.
