Smedjan (the smithy) at Gamla (old) Östanvik is today used as a holiday cottage. The name of the house indicates that this was originally a house used for blacksmith work. It is placed at a safe distance from other buildings at the farm. During repairs work several blocks were taken out of the house. Most of them had been roughly treated by the house longhorn beetle.
Together with Ulf and Lars Häggström, I selected nine blocks from which we cut off pieces from which I considered it reasonable to get datable samples. On six of these blocks, I could later measure ring width series which were going out to a wane edge (bark edge). When crossdating such a series I could thus find out which year that tree was cut in the forest.
Only two of the nine blocks were in a reasonably good condition - both of them were collected from the inner parts of the house. One of them (No 2, see left picture below) which was cut to a quadratic cross-section could later be dated to 1876.
Measuring and crossdating the samples
The samples were first measured with CooRecorder and then crossdated with CDendro, see www.cybis.se/forfun/dendro
For most of the blocks more than one radius have been measured and a mean value ring width curve then created for the tree.
Block 1, 4, 5, 6, 7 och 9 make up a group from six trees where the ring width curves all match to each other. A mean value curve created from this group has been crossdated to the reference curve for Nämdö, ITRDB:swed302.rwl (see reference 1).
When crossdating a sample curve towards a reference curve, the sample curve is placed along the reference curve. A correlation value is then calculated telling how well the two curves match at this position. Then the sample curve is shifted one year along the reference curve and another correlation value is calculated. I.e. for every possible overlapping position, a correlation value is calculated. In the diagram below these values are shown as piles side by side. There is only one high pile corresponding to the year 1822 with the correlation values CorrC=0.74, T=15 which implies a trustable dating of the sample curve. For more information on this methodology, please see ref 2.
The diagram shows that four of the trees were cut in the winter 1822/1823. One was cut the previous year. Because of damage by the house longhorn beetle the cutting year of tree number 4, 5 and 7 could not be established. The youngest block (No 2) was cut in the winter 1876/1877. That block has only 49 year rings, though the crossdating gives a high enough T-value above 6 towards the ITRDB:swed302 reference. So this sample as well as the others can be considered as properly crossdated! According to the Häggström brothers, this youngest block was part of the roof construction.
From this investigation, I guess that the house was built in 1823. Possibly the roof was replaced by a tile roof in 1877 which required some reinforcement work when block number 2 was built into the house.
From this small investigation, I cannot exclude the possibility that the house was moved to its current place in 1877. Though this does not look very probable as we only found one single block from 1876. If the house had been remounted in 1877, we had probaly found more wood from that late date because more old wood had been replaced by fresh wood.
A mean value from all samples gives a correlation coefficient as high as above 0.70 towards the Nämdö reference curve. (As a comparison, the correlation towards a curve from nearby Saltsjöbaden on the mainland gives a corrCoeff = 0.60.) Therefore we can assume that the trees for this house were felt at Nämdö or nearby.
Ring width data is available in Namdo_Smedjan.rwl together with comments that are readable by CDendro.
Saltsjöbaden May 14 2013, updated June 11 2014.