This blog post is part of a series: you may want to read Part 1, Part 2, Part 3, Part 4, and Part 5 first.
Having reconstructed the process and produced an analogue plot of the final grid arrangement, I was ready to replicate the process ‘on the digital ground’ within a Geographical Information System (GIS).
My starting point was not Keiller’s A-B-C baseline laid out across the line of the Avenue (discussed in Blog post 2) but instead the two points A and B where the sides of Keiller’s corridor intersected the fence that closed off the southern end of the field. Whilst the material of the fence has undoubtedly been replaced since 1934, its line is still in position, as evidenced by a marked lynchet. Keiller had recorded a measurement from its junction with the road to point B, and the distance from B to A could be calculated thanks to Pythagoras. This is exactly what I needed – recognisable 1934 locations that were locatable today.
Using these known points to light the blue touchpaper, I then converted Keiller’s meticulously measured distances into metric units and used the GIS to recreate his survey. I did this in two blocks, either side of the 6 degree pivot marked by the Triangle of Correction.
Quality control came in the form of a pair of measurements Keiller took from the midpoint of cutting VIII (at a point he called ‘E’ – see below) along a line perpendicular to the long axis of the grid.
He extended this line until it reached the fence lines bordering the Avenue field to either side and made a record of the distances. When this line was recreated digitally to Keiller’s measurements the endpoints fell within 7cm of the fence-line to the west and 20cm to the east.
The final check was a simple one: did the end of the grid fall within the northernmost E-W boundary fence? I was mindful that the Avenue line ran up slope for a portion of its course, which meant that Keiller’s chained measurements were not taken on perfectly flat ground (as recorded on the modern map). To put this another way, at times he was measuring slope distances instead of horizontal distances (i.e. the hypotenuse rather than adjacent). Yet I was plotting his measurements as though they were taken on flat ground. Over the 450m or so total length of the grid this could result in stretching.
Using height data derived from LiDAR (and three cups of strong coffee), I measured the rise and fall between each pair of 100’ grid points. I then converted the slope distances into horizontal distances to determine how much stretch had potentially occurred. Across the full 450m this amounted to only 17cm. Stretch was not an issue. Although the fence here had long gone, the lynchet was preserved in the LiDAR data and this could be used to recreate the fence line. Plotting the cuttings against this showed a good fit.
Was I happy with +/- 20cm? After 89 years I certainly was. Especially given the lack of fixed reference points between 1934 and today. Given the high likelihood that there had been some degree of movement in the precise placement of the fences, the slope effects, not to mention the inherent precision and accuracy of the Ordnance Survey digital data I was fixing the grids against, this was perfectly acceptable.
So… almost done. In the final blog post I will look at the final steps that were taken in order to create these digital maps of the 1934 cuttings.
Final blog coming soon!