The sea is visible (on clear days) from over 77% of the Dartmoor stone rows*. This may come as no surprise since the moor is situated on a relatively narrow peninsula and views of the sea are therefore likely to be fairly common. However perhaps of more significance is the fact that 30% of all the rows are situated in locations where only part of the row has views of the sea. It is also worth mentioning that a significant number of rows are positioned on the very edge of sea view visibility. The point at which the sea view appears/disappears is precise. One moment the sea is visible, but with a step in one direction it disappears and a step in another it grows. Of course the situation is much more complicated than this and in reality this sea view interface is linear in character and forms an invisible line meandering across the landscape. From any spot on this interface the sea would just be visible. This line is hereafter referred to as the “sea view interface” and is defined as the line separating areas with sea views from those without sea views.
Figure 1. Distribution of Dartmoor stone rows with known orientation and length. The blue stars indicate the position of rows with views of the sea and the red stars those with no sea view. (Base: Google Maps).
The possibility of visual links with the sea was first identified at Bancbryn, where part of the stone row was aligned precisely with Hartland Point in Devon. Indeed the detailed layout of the upper length of the row appeared to take full cognisance of the distant headland. Following this discovery research was directed towards establishing whether other rows had visual links with the sea. A large number of articles on separate sites mainly on Dartmoor have confirmed that visual links with the sea are relatively common, but not universal. The sea view link cannot therefore be the whole solution, but does provide a potentially useful avenue to progress research into the function and purpose of these enigmatic monuments.
Analysis of the view shed from the 81 stone rows on Dartmoor for which information on their length and orientation is known has produced the following result:
Figure 2. Pie chart highlighting the character of the sea views from Dartmoor stone rows
Nearly 50% of the Dartmoor rows have sea views along their entire lengths whilst a further 30% are situated across the sea view interface. In total 77% of the rows have a sea view of one sort or another. This of course means that 23% have no sea view whatsoever and in some places such as Trowlesworthy 2 and Hart Tor, South both rows are situated just beyond the sea view interface. Views of the sea cannot therefore represent the whole answer and at this stage we cannot even be sure whether we can read any significance into these figures.
The 77% sea view figure seems extra-ordinarily high as does the 30% sea view interface figure and therefore worth investigating further. To do this a random series of rows were generated using the RANDBETWEEN function in Excel.
The first stage was to create an Ordnance Survey Easting. This was achieved by entering =RANDBETWEEN(49200,85000). The cell containing this formula was then copied and pasted into 149 cells below. To stabilise these numbers all the cells containing these numbers were copied and then pasted as values only on top of the original cells. If you do not do this the numbers will keep changing. This process was then repeated for the Northings with the formula =RANDBETWEEN(56000,95000) being used. The numbers were then transferred to a GIS system (Quo) and the points outside the National Park boundary removed from the spread sheet except for the area around Lee Moor and Crownhill Down which were historically part of Dartmoor. The upper 81 rows were then retained with the remainder being deleted. This exercise provided 81 random points. The length and orientation element of the random dataset was then provided by manually adding the information from the actual rows onto the GIS mapping. The real rows were sorted alphabetically and the length and orientation information allocated in descending numerical order to the random points. Information on orientation is held on the spread sheet as a figure between 0 and 180 degrees and this was used to create the random rows. Starting from each random point a straight line was drawn at the appropriate angle for the required distance. Each random row therefore shares the dimensions of a real row. The view shed from points along each random row were then created using Heywhatsthat.com (using the 5ft above ground level setting) and the sea view results recorded on the spread sheet alongside each random row. Each row was allocated to one of six categories:
- Sea view from whole length of row
- Sea view from lower length of row only
- Sea view from upper length of row only
- Sea view from middle length of row only
- Sea view from ends of row only
- No sea view
These rows are hereafter referred to as “random rows”. The same view shed exercise was carried out for the real rows providing two compatible data-sets for comparison. The details concerning the character of the sea views were used to create the pie charts and maps.
Figure 3. Distribution of Dartmoor “random stone rows”. The blue stars indicate the position of random rows with views of the sea and the red stars those with no sea view. (Base: Google Maps).
Figure 4. Pie chart highlighting the character of the sea views from the Random Dartmoor stone rows
This chart highlights significant differences with the real rows. For ease of comparison both are presented together below:
Figure 5. Pie charts highlighting the character of sea views between random and real rows.
In broad terms the sea is visible from 77% of the real rows as opposed to 34% of the random rows. This difference strongly suggests that views of the sea played a part in the siting choice and this suggestion is further strengthened by the considerable difference between the real rows built across the sea view interface (30%) compared to the random rows where only 6% were found in this position. This work provides broad support for the idea that many rows were positioned taking cognisance of the sea and the particularly high number of rows built across the sea view interface is likely to represent an important insight.
The spatial character of the differences between the random and real rows can be most easily appreciated by looking at a series of distribution maps looking at the six different characteristics.
Figure 6. Distribution of rows from which the sea is visible along the entire length. (Base: Google Maps).
The majority of the real rows with whole sea views are found in the south western part of the moor and their distribution is significantly more clustered than that for the random rows.
Figure 7. Distribution of rows from which the sea is visible only from the lower end. (Base: Google Maps).
Only the rows at Hart Tor (North) and Leeden Tor have views of the sea from only their lower ends. Both rows are built across the sea view interface. These rows are inter-visible and are situated within the Meavy Valley. No random rows with this characteristic were identified.
Figure 8. Distribution of rows from which the sea is visible only from the upper end. (Base: Google Maps).
Twenty one rows are positioned across the sea view interface in a way that ensures that only the upper part of the row has sea views. By comparison only four of the random rows shared this characteristic. This difference strongly suggests that the sea view interface was often a preferred location for real rows.
Figure 9. Distribution of rows from which the sea is visible only from the middle length. (Base: Google Maps).
No real rows with sea view visibility limited to the middle length of the row were identified. This is interesting and likely to be significant since one of the random rows had this characteristic. This implies that the view of the sea from the top or bottom was sought after.
Figure 10. Distribution of rows from which the sea is visible only from the both ends but not in the middle. (Base: Google Maps).
The central length of the Upper Erme row has no sea views along much of its length but the sea is visible at either end. It is interesting that this very long row leads through a landscape with no sea views, but towards both ends the sea suddenly appears.
Figure 11. Distribution of rows with no sea view. (Base: Google Maps).
There is a marked difference between these two distributions. Most of the real rows with no sea views are in the central part of the moor whilst the random rows are more evenly distributed. This implies that the row builders in parts of the moor where sea views are available generally sited their rows with views of the sea and indeed in many instances ensured that the sea was revealed along the length of the row.
Mapping depicting the areas of Dartmoor from which the sea can be viewed is available at DATA.GOV.UK. The resolution of the underlying dataset is rather low with readings taken on a 500m grid. However at the overall Dartmoor level the map produces an insight into the areas where the sea can be viewed from. The extent of Dartmoor sea views in shown in Figure 12 below which highlights an obvious correlation between stone rows and sea views. Only about 30% of the moor has views of the sea and yet 77% of the rows are built within these areas. This confirms the separate findings presented above that the distribution of the real rows has a correlation with sea views. The rows were clearly more likely to have been built in areas with views of the sea. A second point to be taken from this map is that it validates the random row figures and explains why 66% of the random rows have no sea view. This figure is wholly appropriate given that only about 30% of the moor has sea views. It is undoubtedly significant that the largest blocks of land with no sea views also contain no stone rows. It is also interesting that most of the rows are situated towards the edges of the sea view areas. The case for a correlation between sea views and rows at both the general and individual level is strong and this in turn indicates that views are likely to have been a significant impact on the choice of location.
Figure 12. Distribution of Dartmoor stone rows with known orientation and length. The blue stars indicate the position of rows with views of the sea and the red stars those with no sea view. The areas with sea views are shown by solid yellow (Base: Google Maps and Marine Management Organisation).
These distributions maps make it clear that views of the sea of whatever character were not the only criteria employed in the choice of site and therefore that other factors must be at play. Most of the rows also have definable visual links to other landmarks and it is therefore best to see the rows as having been positioned to reference a range of landmarks, of which the most obvious is the sea. It is very likely that different places of significance were incorporated and the obvious correlation between the real rows and sea views provides strong evidence that the row builders were interested in visual links with significant places in their world. At an individual row level it has been possible to identify possible visual links of significance and the case for the rows being special places providing particular views of other special places is powerful. The work looking at the differences between real and random rows provides substance to the idea that the row builders were interested in sea views and this in turn implies that such views were important to them and influenced their choice of location. The obvious step of suggesting that other views may have also been of interest is logical and likely and is supported by observations at the real rows themselves.
Clearly other reasons could be responsible for the observed visual links, although this work has at least indicated that their distribution is likely to be far from random and that the most plausible explanation is that amongst other things, particular views of the sea were important to the row builders.
A similar conclusion was independently reached by Neil Mortimer who noted that:
“I began to notice that monuments, and stone rows particularly, on the southern edge of Dartmoor had views of the distant coast. Some of these views are spectacular and obvious; other are subtle and little more than glimpses through the landscape.” (Mortimer, 2013,232).
So at the same time the visual links with the sea were being identified at Bancbryn, Mortimer was noticing similar relationships on Dartmoor.
The particular reasons why each stone row was built precisely where it was will inevitably remain uncertain, although with a developing understanding of the importance of visual links it is hoped that these places will become a little less enigmatic.
* For the purposes of this exercise only the rows where the length and orientation of the row is known have been used. Six rows where this information is not available are excluded. As work proceeds the information used in this analysis will inevitably be enhanced. This article therefore should be considered as a draft to encourage discussion and the development of ideas and will be updated in the future as research and fieldwork inevitably alters the dataset upon which it is based.
I would like to thank Bill Radcliffe, Helen Woodley and Sophie Smith for their considerable input into this article. Without their help it could not have been written.
Mortimer, N., 2013, ‘From Moor to Shore: Re-Visioning Dartmoor’s Stone Rows’, Time and Mind, 6:2, 231-240.