by William A. Flynt, Historic Deerfield, Inc. and Myron O. Stachiw
Presented at the VAF Annual Conference, Salt Lake City, Utah June 3, 2017
During the past several years Myron Stachiw and William Flynt have been engaged in research related to a historic structures report of the Hancock-Mitchell House, a historic dwelling house in Chilmark, on Martha’s Vineyard, Massachusetts. This project has revealed the complex and often conflicting relationships that exists between local historic heritage and historical fact about buildings, places, and events. Local knowledge and previous research conducted on the property identified a portion of the building as a 17th century structure, associated with the noted Indian missionary Thomas Mayhew, Jr., or with his son John, also a missionary. The physical and documentary research conducted by Myron Stachiw for the historic structures report on the building, assisted by a dendrochronology study conducted by William Flynt of Historic Deerfield, dated the building as having been constructed nearly a century later than believed by everyone, from the previous researchers to the local historical commission, the island’s preservation trust, and the property owner (Figure 1).
The goal of the initial dendrochronology study undertaken by Bill Flynt in 2013 was to ascertain the dates of the three major phases of the building’s construction history (Figure 2). Phase 1 and 2 oak samples could not be successfully dated due to the lack of local oak masters and their inability to align with dated mainland masters, but it was possible to determine that the age difference between Phases 1 and 2 was 42 years. All physical and documentary evidence of the Phase 2 construction and finishes suggested a date of construction between the 1790s and 1810s, resulting in a date of c. 1750s-60s for the Phase 1 portion of the present building. The hemlock used in the framing of the Phase 3 rear ell was able to be successfully correlated with mainland hemlock masters as having been felled in 1836.
In 2015 a number of timbers were replaced during the restoration of the structure, and more of the house framing was exposed. This lead to a request for further testing, as the conclusions resulting from the HSR and initial dendrochronology study was not accepted by the client and community as conclusive or reliable. If several other early island buildings could be sampled, the chances of aligning all buildings and getting one or more to successfully date against mainland masters would increase. Three homeowners of early houses agreed to participate, and in November of 2015 additional sampling was undertaken at the three buildings and at the Hancock-Mitchell house (Figure 3).
Most of the oak timbers sampled were successfully aligned to create a floating master - one where age differences can be established between samples, but not assigned specific calendar dates. Comparing the oak data against a number of mainland oak masters, certain samples from each of the structures appeared to align well with specific dates that were offset in conformance with what the floating master displayed. This revealed that the Butler-Strock house timbers were felled after the 1743 growing season ceased, the Parsonage' s timbers were felled primarily in the winter of 1744/5, and the Phase 2 timbers of the Look-Horwitz house were harvested in the winter of 1786/7 (Figure 4). At the Hancock-Mitchell house, the new data allowed dating the felling of the Phase 1 timbers to the winter of 1759, with a few felled a year earlier, while Phase 2 timbers came down during 1800 and 1801.
While Bill Flynt was 95% confident that the results were accurate, a book titled Pioneer Houses of Martha's Vineyard was published by one of the previous researchers who dated the Hancock-Mitchell house to the 17th c. He continued to maintain the 17th c. date for the house and also claimed similar 17th century dates for two of the other three houses. This caused Bill to look for yet another way to confirm his results.
Paul Krusic, the dendrochronologist responsible for getting Flynt involved in this line of work and who remains his mentor, suggested it might be possible to use Accelerator Mass Spectrometry Carbon 14 (AMS14C) testing on one or more samples to see if it could help confirm (or refute) the provisional dating as being correct. With AMS Carbon 14 testing it is possible to narrow down a date to plus/minus 20 years or less.
Bill Flynt worked with Todd Lange of the University of Arizona AMS Laboratory, who noted that the carbon calibration curve for the northern hemisphere declined throughout the 17th century before reversing course in the early 18th century (Figure 5). This offered the possibility for a two-part analysis that had the potential to eliminate the plus/minus 20 year range, but had never been attempted by the lab. The idea is to select a sample from a ring suspected as being at a unique position (date) on the curve and, if confirmed, to then sample a ring at a second transition point on the curve and hope the results align with expectations. Should the two tests produce results as expected, this would indicate the sample is correctly dated as determined by the dendrochronology analysis.
Working with the Hancock-Mitchell Phase 1 sample that tentatively dated the timber to having been felled in 1759, material was extracted from the ring assumed to date to 1715, as it should correlate with the lowest part of the early 18th century “dip” in the calibration curve. Figure 6 depicts the results of the sample analysis. Of importance here are the gray shaded areas indicating where the sample's newly-calibrated Carbon 14 content falls along the calibration curve. Three areas of strength are noted, one in the early 18th century, one in the early 19th century, and one at the turn of the 20th century. Knowing which Phase 1 timber the sample came from, it is safe to dismiss the 20th century spike and most likely the early 19th century show of strength, leaving only the early 18th century alignment, which actually peaks at just about 1715, a promising result.
A second test was initiated using material from the ring 25 years younger (assumed to be 1740) as the Carbon 14 calibration curve rises to 1740 and then meanders for a time. Should the sample reflect this increase in Carbon 14, it would then negate the 19th and 20th century results noted on this report, as in neither case is there as dramatic an increase over the ensuing 25 years. Figure 7 displays the results of the second analysis indicating the increased amounts of Carbon 14 in the sample. While there are numerous areas along the Carbon 14 calibration curve where this sample could date to, the strong peak at 1740 is the only place where these results correlate with those of the first test. Thus, the AMS Carbon 14 analyses confirm the correct dating of the sample, which in turn validates the Martha's Vineyard Oak master, and the dating of the houses associated with it.
While dendrochronological methods provide absolute felling dates most of the time, when results are less clear, two- pronged AMS Carbon-14 analysis provides us with a remarkable new tool to significantly improve the accuracy of dating historic buildings.