2013-04-15

On Historical Maps and Rectification


The rectification of historical maps is a commonly used tool for understanding landscapes and archaeological sites. It is often used by contract archaeology to determine the location of potentially interesting features. By landscape archaeologists rectification is used to make landscape analysis in order to recognize certain changes. But while this tool is effective and easy to use in most GIS softwares it is by many geographers considered to shallow and inaccurate. The rectification often stretches the original map to a point were the historical surveyors’ measurements are no longer valid. It is true that historical maps may be inaccurate, but most of times they are surprisingly well made, as discussed below. Therefore it should be considered whether stretching of original maps is a good way to work or not. The accuracy of the surveyor must always be taken into account. Also, rectification does not often promote an actual understanding of the map, it merely tries to force common points without too much thought. I have chosen to take the perspective of the geographer in this, and the method described below shows another good way of working with historical maps and in turn rectification. Most of this is based on the methodology of the project I’m currently working on dealing with 17th century maps from Sweden. If you want to know more please visit http://www.riksarkivet.se/default.aspx?id=21561&refid=22519 or send me an email!

   First we must recognize the parts of the map that have not changed for 300 years or so. Many times roads are used for rectification of historical maps, but these are in my opinion quite uncertain elements and it is possible that they have changed a lot during the centuries. This also applies for humanly created features like farmsteads, bridges and sometimes also large manors and churches. The only humanly created feature that can be used as points of rectification with a quite good accuracy are the administrative borders found in historical maps. These are quite surprisingly often still used today and although some changes have been made it is usually not a very hard task to recognize parts that are unchanged. Otherwise, it is usually the natural topography that gives us a hint of common points. Mountains and rocky parts of land were usually not cultivated in the 17th century, and are commonly still not cultivated today. In the map of Hammarby, we can see such a large patch of land just northwest of the village itself. Also, to the direct southeast of the village there is a perhaps more useful round patch in the farming fields (marked by green ring, bottom left corner of present day map). Just south of the label Backgården there is another fairly certain point (marked by the blue ring).


A 17th century map of Hammarby in Uppsala county. The village is at least medieval and fairly large, though its exact origin remains quite uncertain.




A modern day economic map of the same area. The positions of the farmsteads in the historical map are marked with red points

   It is surprising how accurate the some 17th century Swedish surveyors actually were. It is usually said that they were the most accurate with the actual fields and within the boundaries of the farmsteads, but in measuring in the actual map itself it is proved that their abilities went far beyond that. Of course, this depended on the surveyor himself, as some were much more accurate and detailed than others.
As a result of this relatively high accuracy it is usually possible to use the scale bar (marked with a red ring in the picture) found in the historical map itself to make measurements. In the 17th century maps of Sweden, the units are measured in “alnar”. 1 “aln” is the same as 59.3 centimetres and a good way to start is by converting the scale bar into modern units. This can be done easily with the program DjVu viewer (http://www.djvuviewer.com/) if your file is in djvu format. Otherwise, though perhaps more complicated, you can probably reach the same result using your GIS software or CAD. In doing this you can calculate the relative accuracy of your particular surveyor by measuring distances between topographical features and boundaries, both in the historical and the present-day map.

   After this procedure you can use the common points in topography and administration described above to triangulate the position of features of interest in the historical map. Usually it is enough to take two measurements in the historical map, from the common point in the landscape to the feature. When these two measurements correlate from the same points in the present-day map you are usually quite close to your feature.

   To give support to the position of your features it is usually a good thing to consult your country’s record of monuments and sites. In Sweden, the online database called FMIS is a good resource for validating your results. Also, as it is always good to use multiple sources, you should also consult more recent historical maps. In the case of Hammarby I used an economic map from the 19th and early 20th centuries called “Häradskartan” were some of the old features still remain. Fields and topography are also more easily recognized. These historical yet quite recent maps are usually considered a bridging point between older and present day maps.

How these points can be used in making a rectification must still await a complete evaluation, but they should at least provide more valid information on the position of the mapped features. In this particular map, for an instance, I have only put a few points in the centre, which would not be a good base for rectification of the complete map. This is rather another way of recognizing common features in the maps. In order to make a good rectification I would need to use the whole of the quite extensive map, which is not possible at the moment, sadly. I will come back with a rectification result when I have gathered enough data to do so. Until then, I hope that my geographers perspective have proved inspiring!

NOTE:
All maps taken from http://www.lantmateriet.se/

2 kommentarer:

  1. Great post!

    Whereas I agree with everything you said, I think that some degree of caution needs to be exercised even with natural topography. Bodies of water, especially rivers, have the tendency to fluxate—sometimes even dramatically. Furthermore, they are such a visible and recognisable feature in the landscape that it is difficult to avoid using them as landmarks. It is also important to recognise that, yes, mounds of rocks and mountains weren't necessarily cultivated in the 1700s, but today mechanised farming certainly allows for the removal of such rocky landscapes if necessary. As for mountains, mining frequently turns them into craters—one just has to look eastwards of Lund, Sweden to see this {the first link shows an active quarry, the last two show old quarries that are now lakes):

    Modern quarry at Hardeberga (https://maps.google.com/maps/place?q=55.702258,13.2932)

    Dalby Stenbrott (https://maps.google.com/maps/place?q=55.65919,13.402505)

    Billebjer (https://maps.google.com/maps/place?q=55.68953,13.317297)

    Of course as you have instructed the reader, this is why it is absolutely crucial to be sure that you reference multiple maps sources. Only in this way can a historical and modern maps be bridged.

    I am looking forward to seeing the rectification results once you have obtained more data!

    SvaraRadera
  2. Thank you for your comment Justin!
    On the topic of rivers, I totally agree with you. In flat areas such as the one in the map above, rivers tend to meander heavily as the flowing speed is so low. It is sometimes possible with some basic knowledge of river meandering patterns to discern the history of changes by merely comparing the two maps, but still rivers should be used as reference points only in lack of other more valid sources. In areas with a more steep topography, rivers tend to cut downwards instead of sideways and in those cases they can maybe be used more accurately. Sadly though, those areas were almost never mapped in the past, at least not with the same precision.
    The mountains are in my opinion a lesser problem, as it almost always becomes clear what mountains that have changed and how. I would guess that stone quarrying constitutes a bigger problem in Skåne where the farming fields nowadays are quite extensive. In more forested areas like Uppland (which I am working with at the moment) the extent of farmed land is almost always very much the same as in the 17th century, and the "mountains" have rarely been removed. Here, only the low lying valleys between the mountains have been used for farming. These valleys are filled with sediments from the Ice Age and usually creates a good soil environment for farming.

    SvaraRadera