Braathen Dendrokronologiska Undersökningar

Garde church, Gotland

Garde (Sv.K 145). The parish church of Garde in the south-eastern part of Gotland contains the oldest pieces of wood that I have dated from any other stone church. The roof structure is made of pine except for the shingles which are made of oak.

The building is described in detail in E. Lagerlöf, Garde kyrka. The book contains 136 illustrations, which are commented in english. On page 343 and 344 is an english summary of the swedish text.

The belfry is joined with bonded dressed stones to the nave. This tells us that these two parts of the building are contemporary with each other, but the belfry has later been made higher, probably when the nave got a gothic roof.

On top of the western part of the nave the main part of the romanesque roof is preserved. The planking of the roof is horizontally arranged and fixed to 17 pairs of main rafters by iron nails. The planks are wedge-formed at the edges so that they cover the edges of their lower neighbours and would prevent raindrop if necessary. There has not been possible to find by certainty waney edges among these planks. Nine samples have been taken from the planks on spots where earlier these have been damaged. The dates of the outermost examined annual rings are distributed as is shown in the following table:

Identification Dates
31 the nave, planking of the roof 1116
2 1114
30 1108
19 1104
3 1092
1 1051
20 1050
4 1037
5 Not dated.

The distribution of the dates of the outermost annual rings show the importance of taking a sequence of samples in a set-up where the timber seems to lack waney edge. Let us assume that we had taken one sample only and that we at random had picked up the sample which gives the date 1037. That date does not give us wanted information. The first four dates are rather compact, and we may assume that the timbers originate from trees which were felled about the same year, but we only know that the earliest possible felling year is the winter season 1116 - 1117. We also know that it is rare to find pines which have grown for more than 400 years. Sample no 31 contains 155 narrow annual rings and we can assume an upper limit for the possible felling year to be in the middle of the fourteenth century. Still the dates in our sequence are not useful unless we can combine them with other facts.

Let us examine how planks were cut from a tree trunk by means of axes and wedges. When cleaving, it was favourable to follow the direction of the wood fibres. That is why straight grown trees were selected. A study of medieval timbers shows that when forming straight lines and surfaces the surface of the trunk was followed as far as possible. For that reason it is important to study the direction of the pattern of the annual rings in a timber in relation to the edges.


We return to the first four dates in our sequence. The difference between 1116 and 1092 is 24 years. we do not know how thick layer of wood was removed from the timber up to where the youngest dated annual ring was reached. That layer might as well correspond to 24 years and we reach a felling year that is about 1140 or the middle of the twelfth century. An examination through microscope of the youngest annual ring gave no sign of cutting from an egg tool. That annual ring might be the closest one to bark, and a study of the edge of the plank which is protected by an adjacent plank gave the impression of waney edge but we do not know.

The three older dates in the sequence differ with 64, 65 and 79 years from the youngest date 1116. Sample no 4 with the date 1037 has a normal ring width of about 1 mm. If this ring width broadly was unchanged in that part which has been cut away we get as result a thickness of 80 mm of the removed wood which is much. Therefore there is reason to suspect that the timbers with the three older dates may have had an earlier felling year, or at least, we can not be certain that they are associated with the still unknown felling year, which is connected to the youngest date 1116.

The discussions presented above are intended to show the importance of taking many samples from a set-up, and that many problems may be involved in the attempt to interpret the result of the dating. In our case we want as much relevant information as possible, e.g. what we know about time-dependant details in the building.

So far our conclusion from the presented 8 dates is that the felling year for the trees is earliest 1116 without a defined upper limit, but where the middle of the fourteenth century has a low degree of probability.

The above mentioned roof was in the gothic period functionally replaced by a roof with a higher rising which explains why the cover of the romanesque roof has been protected from weathering.

It is not possible to determine when the gothic roof was erected because all the timber from that period was removed at a repair in 1968. The cover of the romanesque roof is partly preserved and consists of exceptionally big shingles made of oak. At my visit in 1982 the floor was strewn with broken remains of oak shingles from which I cut samples and put the rest to the place where the piece of wood was found.


Shingles of oak normally are cut from heavy trunks so that the surface of the shingle follow the direction of the wood rays. Trunks with a wide diameter allow two shingles to be cut along the radius which explains why the distribution of dates from shingles, contemporary with each other, is widely spread. The sapwood normally is removed, but in fresh wood the boundary between heartwood and sapwood is not always conspicuous, and sometimes part of the sapwood may still be preserved in that part of the shingle which is protected by its overlapping neighbour.

Referring to the distribution of the number of annual rings in sapwood (See Sapwood and waney edge) where in the middle of Sweden the mean value of the number of annual rings in sapwood is 14 with a standard deviation of +3, -2 years, we get for samples no 11 and no 23 the felling years in the periods 1123 - 1128 and 1120 - 1125 respectively. Usually we get the same limits of the calculated periods of felling for trees which we know are felled at the same time. As result we get that the oak shingles were made in the period 1120 - 1128 and that the building of the nave and the romanesque belfry were complete at that time.