You are on page 1of 19

Contents

1. Summary

2. Introduction

3. Method

4. Results

5. Conclusions

6. Acknowledgements

7. Illustrations
7.1 Village of Cramond, Cramond House and Tower in relation to fort and survey.
7.2 Surveyed area grid, linear array lines and magnetometry areas.
7.3 Area survey resistive printout.
7.4 Interpretation of area resistive survey.
7.5 Linear array section printouts.
7.6 Generic Mapping Tool. Shaded relief plot, gradient data.
7.7 Magmap 2000 shaded relief plot (Squares G 18,19,22 and 23)

8. References
8.1 Historic Scotland letters of 11 July and 20 November 2002 Ref. AMH/2526/1/1
8.2 Historic Cramond Management and Interpretation Plan. Marta McGlynn Assoc.
8.3 Parish of Cramond. John P. Wood. (1794)
8.4 The House. Jean M. Crichton in Cramond (Cramond Heritage Trust 1996)
8.5 Plan of Cramond. Property of the Rt. Hon Lady Torphichen.
Thomas Bauchop (Surveyor) 1815
8.6 Drift Geology Sheet NT 17NE Drift 1: 10,000
8.7 Solid Geology Sheet NT 17 NE Solid 1:10,000
8.8 Archaeological Geophysics at the Roman Vicus area of Cramond. December 2004
Greg Macdonald. University of Edinburgh School of Geosciences
8.9 Archaeological Investigations in the Walled Garden, Cramond, Edinburgh
H M D Jones EAFS Geophysics Occ. Paper No. 1
8.10 Excavation of Roman sites at Cramond. Nicholas Holmes.
8.11 A design for the improvement of Cramond, the seat of Sir John Inglis Bart
Thomas White 1797
EDINBURGH ARCHAEOLOGICAL F E L D SOCIETY

Geophysical Investigation in the Parkland to the East


of Cramond House Edinburgh

1. Summary

An area ground resistance survey of the parkland to the east of Cramond House
showed, in the printout of the data, a number of high and low resistance lines parallel
to the tree avenue that is aligned on Berwick Law. The complexity of these lines, with
some at right angles, suggests that a formal garden was created at the time of the
building of the first house in 1680 There are also a series of lines that cross the tree
avenue at an angle of about 40 degrees and appear to be at right angles to the Roman
road that runs to the south of the surveyed area. These could relate to the Roman vicus
some parts of which are known to lie to the east of the fort. Three linear array resistive
measurements were made across the approximately E-W low resistance that shows on
the area printout on its northern edge. Aligning with this low resistance the linear
array printouts show what appears to be a ditch running along the edge of the old
raised beach level. The slight high resistance on the north side could represent upcast
from the ditch.
Two less regularly shaped high resistance features could be gravel beds related to the
formation of the old raised beach The largest is about 20m. fiom, and parallel to, the
raised beach edge and appears in a hrther linear array section as 5m, thick whereas
that which is 50m. from the raised beach is not greater than 2m. in depth and may be
archaeological rather than quaternary geological in origin.
The magnetometry survey showed a number of linear features that were roughly
parallel either to the tree avenue or to the "Roman" features. One of the magnetic
anomalies did not align with either high or low resistances in the area resistance plot
and the three resistive linear array measurements across it were inconclusive.

2. Introduction

The Society requested, in July 2002, permission from Historic Scotland to conduct
geophysical surveys in four places within the Scheduled Monument area around the
Roman fort at Cramond. Permission was granted in November 2002 by letter
(Ref.8.1) to a modification to the original request increasing the area of survey $0 the
east of Cramond House. This area falls within that commented upon in the June 2003
report by McGlynn (Ref.8.2).In this report under the heading "The early formal
landscape (1622-89)" the report states "below ground archaeology of the
accompanying 17" - 18" century formal garden or parterre remains to be verified.
There is also a more general statement that c'unrecordedremains are very likely to
survive buried in the Scheduled Area". Geophysical survey is suggested to "establish
evidence of any formal garden associated with the 1680 house". The permission,
already in place at the time the report was published, allowed the Society to start
planning the area ground resistance survey immediately.
Cramond was in the possession of the Bishops of Dunkeld from at least 117314 when
Richard de Prebenda, the second bishop, is recorded by Wood (Ref.8.3), as dying
there; he also records that, in 1574, "the first Protestant bishop thought proper to
alienate the town and lands of Cramond to Archibald Douglas of Kilspindie". Having
passed through two other hands it was bought in 1622 by "James Inglis, a merchant in
Edinburgh". James died in 1637 and was succeeded by his son John (16 11- 1684) who
built the oldest part of the house in about 1680. His son, another James, was created a
baronet in 1687 and died in 1689. It was his grandson, Sir Adam Inglis (1714-1 772),
who added the east face to the house in the year of his death. However Crichton
(Ref.8.4)states that the eastern wing was added for Sir Adam and his brother Sir John
about1778. It is McGlynn (Ref.8.2) that records that the original house builder
"develops a formal designed landscape" and may have been responsible for the beech
avenue to the east of the house. The extension by Sir John and Sir Adam developed "a
more formal landscape which appears to be shown in Bauchop's 1815 plan" (Ref.8.5).

The parkland to the east of Cramond House lies above (to the south of) the old raised
beach level with the centre of the survey approximately at NGR NT 192 769. The
remains of a tree avenue extend eastwards across the parkland framing a view of
North Benvick Law at a distance of some 37 km. from the house. The Bauchop estate
plan, of 1815, shows this tree avenue (or its predecessor) and appears to indicate that
a ditch and path follow the contour of the edge of the raised beach eastwards and then
curve to the south on the line of an existing small, rather marshy, burn.

The original broad survey objective was to record and interpret what was to be found
in making the geophysical surveys over the area measuring, at its widest, 100m. N-S
by 120m. E-W. The ground is roughly level until it slopes away on the north side at
the edge of the raised beach. The house lies to the west and the now disused and
overgrown driveway is beyond the fence on the south side of the area. The fact that
the estate plans show no change in the parkland for over 200 years suggested that
Mesolithic, Roman and later features could remain in situ as they did in the trenches
excavated, by the Society, adjacent to the kennels building. The report on this
'excavation is in preparation. The village of Cramond, Cramond House and Tower, the
kennels and their relationship to the Roman fort and the surveyed area are shown in
Illus.7.1.

The initial survey conducted was area ground resistance to a depth of 0.5 to 0.75m.
A total of 9,200 sq, m. was surveyed, one reading being taken in each square. These
readings were taken in the first six months of 2004 in eight separate day sessions.
The area survey was followed in October and November 2004 with resistive linear
array measurements over particular features that showed in the area survey printout.
The information derived from the resistive surveys was then used to decide the area to .
be covered in two magnetometry surveys.
The drift geology of the area (Ref8.6) indicates that the raised beach is about 29ft.
over M.S.L. and the edge of the terrace, after heading east for some 150m.from the
house, is shown as curving to the south along the line of the burn mentioned above.
The raised marine deposits are Devonian with till and boulder clay along the line
where the ground drops away to the north. The solid geology (Ref8.7) shows the area
to be Lower Oilshale Group of the Carboniferous Limestone Series with a fault line
running approximately NNW-SSE about 50m. to the east of Cramond House. To the
south and west of this line is Craigleith sandstone; about 120m. east of Cramond
House a teschenite intrusion, about 50m. wide, runs approximately N-S terminating at
its southern end at the fault line. The depth of topsoil over the marine deposits is not
known but the area survey shows non-linear high resistance areas that could represent
gravel banks and similarly shaped low resistance areas where dune-slack ponds may
have existed some 70m. to the south of the raised beach. Nothing that can be
interpreted as bedrock shows in the ground resistance printout and this is confirmed in
the limited area covered in the 24m. (1 .Om. spacing) linear array sections. When the
electrode spacing was increased to 5.0m., giving a 120m. section, the printout
indicates comparatively low resistance at a depth of 20.0m. suggesting that bedrock
has still not been reached. It should be noted that this long linear array section was
made approximately 100m. from the east face of Cramond House and was thus over
the sedimentary rocks to the west of the teschenite intrusion and the comment on
bedrock depth may only apply in this sedimentary area.

3. Method

The area ground resistance measurements were carried out using a TWCIA mobile
probe frame with the data stored within the unit and later downloaded, via the RS 232
interface, to a computer running the TWCIA software programme. Probe spacing on
the mobile frame is O.5m.and readings were taken at l.Om, intervals over the total area
representing 23 standard 20 by 20m, squares. Dummy readings were inserted where
fencing or trees prevented actual measurements. Two remote probes, one for current
input and one for measurement, were located not less than 20m. from the area to be
surveyed and required repositioning as the survey progressed. Adjustment was made
to.the spacing of the repositioned remote probes to give, if possible, an identical
resistance reading to that recorded when they were in their previous position. Any
minor resistance discrepancy was corrected by the computer "edge matching"
programme.
The linear array ground resistance measurements were made using a Campus
. Geopulse earth resistance meter connected to a line of 25 electrodes spaced initially at
1.Om, intervals. Apparent resistivities were measured using a lap-top computer
running "Image" software by Campus which takes readings using different
combinations of the 25 electrodes. The computer progamme controls into which
electrodes current is injected and between which pairs voltage readings are taken for
the computation of resistivities. The total number of readings taken is 92 on each line
giving a pseudo-section length of 22.0m. By increasing the electrode spacing to 5.0m:
the length of the array line becomes 120.0m. and the depth to which measurements
are taken increases in proportion to the increase in electrode spacing. The increase in
electrode spacing, however, results in a loss of definition of features that are smaller
than the electrode spacing. The linear array measurements that were made over the
edge of the old raised beach level were corrected topographically as there was a 3m. -
change in ground height over the 24m. length. Readings were taken at each electrode
point, using a Watts Quickset level SL 10, and these were fed into the computer to
correct the section printouts. Survey data was processed using a programme
'Res2DinJ' produced by Advanced Sciences Inc. The programme applies an
iterative, smoothness constrained, least squares inversion to the recorded apparent
resistivities to produce a two dimensional resistance model of the sub-surface.
The magnetometry survey was carried out using a G-858 Mag Mapper caesium
vapour magnetometer in a gradiometer configuration. In conjunction with this an
integral satellite navigation equipment recorded the spatial position of each point of
magnetic reading.

Fieldwork was started by the Society with the layout of the 60m. long baseline in the
parkland immediately to the east of the small car park in front of Cramond House on a
line 3 .Om.fromthe flower bed dwarf wall. The origin point was in the extreme SW
corner adjacent to the fence. Eastwards from this line the 20 by 20metre squares for
the ground resistance area survey were planned and the total of 23 squares were laid
out. Resistive linear array measurements and the magnetometry survey could not be
arranged with the University of Edinburgh until the autumn term and these were
completed in October and November 2004.
The layout of the resistive area survey, with grid numbers of the data store, is given in
Illus.7.2 and included are the positions of the linear array lines and the areas covered
by the magnetometry surveys. The positions of linear array lines 1,2,3 and 8 were
specifically chosen to intersect features that appeared to be of particular interest on
the area resistive survey. Similarly the areas covered by the magnetometry were
placed where it was hoped that additional information might clarifl the interpretation
of the area resistance plot. The first magnetic survey was made with readings at 2.0m.
spacing over squares G 11 to 25 and G 29 to 3 1. Readings taken in squares G 8,9 and
10, adjacent to the fence, showed only corrupted readings due to the wire within the
fence and these have not been assessed. Squares G 18,19,22 and 23 were resurveyed
with readings taken at 1.Om spacing to attempt to plot, more accurately, anomalies in
that area. Linear array lines, 5,6, and 7, were surveyed across a 20m. long magnetic
anomaly that angled through this latter area.
The area resistance survey, the associated data processing and the topographical
correction data readings for the resistive linear array measurements were taken by
members of the Society. The linear array and magnetometry surveys were arranged
initially by Dr Bruce Hobbs of the School of Geosciences of the University of
Edinburgh as a student project undertaken by Greg Macdonald; this was latterly
supervised by Dr Roger Scrutton. The data and printouts of these surveys were
produced and provided by Greg Macdonald in his report Ref, 8.8.

4. Results

The resistive area survey, Illus.7.3, shows a considerable number of features that
comprise both high and low resistance lines, some up to 100m. in length, that run .
parallel with the avenue of trees, on a bearing of 79 degrees, in the direction of North
Benvick Law. This, together with some lines at right angles to the tree avenue line,
suggests that a more complicated garden layout predated the parkland. A marked low
resistance line runs, fairly clearly, along the northern side of the area printout with a
narrower high resistance on its N side. The low resistance is about on the break of .
slope at the point where the edge of the old raised beach level is shown in Ref. 8.6.
The linear low resistance curves to the N at the W end of the plot and to the S at the E
end where the parkland "plateau" ends at a marshy burn. The burn is likely to have
been larger in antiquity as Ref, 8.6 shows the raised beach line curving at this same
point. A rather less clear series of resistive features appear to cross the line of the tree
avenue on a bearing of about 40 degrees and could be seen at being at right angles to
the Roman road that leaves Cramond fort on a bearing of 128 degrees. An
interpretation made before the survey was hlly completed is shown in Illus. 7.4.
A total of eight resistive linear array measurements were made, four of which were
aligned to give sections through the linear low resistance on the line of the raised
beach edge (Lines 1,2,3 and 8 in Illus.7.5). It should be noted that in these and the
later linear array measurement printouts, the colours relating to the resistance scale
vary; comparing colours between section prints can thus be misleading. The depth of
the large high resistance in the middle of the area printout and also some indication of
the site geology was derived from the 120m. array line number 4 which, over part of
its length, is co-linear with line 3. The angled magnetic anomaly found in squares G
18 and 22 had three linear array sections made across it, two were across the longer
limb (lines 5 and 6) and the third (line 7) was made at right angles to the previous two
across the less distinct short limb in order to see whether any associated resistive
anomaly could be traced. These three sections were made with the electrodes at 0.5m.
spacing in order to give a more detailed picture of'the section. Due to reducing the
electrode spacing the section depth is limited to 1.6m. No very obvious resistive
feature aligns with the magnetic anomaly.
Analysis of the magnetometry sensor data suggested that the positions recorded by the
G.P.S. were not as accurate as those derived from walking the 1.Om, lanes laid out
within the grid squares. The magnetic anomalies showed up more clearly when the
gradient, the difference between the field measured by the upper and lower caesium
sensors, was plotted in the 1.Om. lanes. Due to a slight misalignment of the sensors
"striping" appeared on printouts where one lane was walked from S to N and the
adjacent lane N to S. A "de-striping" programme corrected this error.
The printout of the large area magnetic survey, with readings at 2.0m. spacing, (Illus
7.6.) shows anomalies orientated both on the "Roman" and tree avenue alignments.
The smaller magnetometry survey, over squares G 18,19,22 and 23 (Illus.7.7),
indicates more clearly an anomaly in G18, that is evident on the larger survey,
together with a rectangular shape, in G22, that is not identifiable on that alignment in
Illus. 7.6. Neither of these anomalies appear to be aligned precisely with the "Roman"
features. However a re-interpretation of the rectangular magnetic anomaly in G22 in
Illus. 7.6 suggests it is co-linear with a resistive interpretation in Illus. 7.4.

5. Conclusions

The linear low resistance, on the area printout, running for about loom, along the
edge of the raised beach with a shorter high resistance on its N side was examined in
section, through square G21 (and extending 4.0m into square G20), by three 24m.
linear array measurements, Illus. 7.5, lines 1,2 and 3. These sections, spaced 5.0m.
apart E-W and taken with electrodes at 1 .Om. spacing, show a 2.0m. wide low
resistance that is approximately 2.5m. deep at the 16.0m. point on the line. The shape
of these suggests a linear ditch. The high resistance line on the N side of the ditch in
the area printout also appears in the sections and may represent upcast from when the
ditch was cut. In comparing these ditch sections with the linear array printout
Fig.6.3.2.in Ref.8.9 a similarity is seen with the low resistance that lies between the
19.0 and 21 .Om.marks.This linear array measurement, made in 2002, was laid out to
cross the inner and outer ditches of the Roman fort to the W of Cramond Church hall.
The inner ditch is known to lie in this 2.0m. wide section. The similarity between the
two sections suggests that a Roman ditch could have run along the edge of the raised
beach level..
Holmes, in Ref. 8.10, states that the Roman ditch beside Cramond Tower (Area VII)
was "running at right angles to a line continuing that of the east ditches of the fort".
Had this ditch continued E on that alignment it would have crossed the area resistive
survey diagonally through squares G 10,13,17,20,24and 29; no indication of a
resistive anomaly on this alignment was found. It is conceivable that the Area VII
ditch was constructed to curve and utilise the natural slope of the raised beach for
defence. A hrther area resistance survey heading E from the Tower and in to the NW
corner of square GI0 (where some dressed stone lies in the fence line) could clarifL
the line. Scaling and superimposing the 1815 Bauchop plan (Ref. 8.5) on Illus.7.1
suggests that the un-named feature, with footpath beside it, is co-linear with the low
resistance line of the ditch. The Bauchop line of the footpath and ditch curves from
beside the tower to be straight for 100m. to the N of the tree avenue and then swings
S. following the edge of the raised beach alongside the burn. After about 60m. on this
line both turn to the SW and ultimately but onto the side of the walled garden. The
low resistances in the extreme SE corners of squares G22 and 33 could represent this
line. In the final 20m, adjacent to the walled garden, the path and ditch cross the line
of the Roman road entering the fort and thus are unlikely to be contemporary with it.
Additionally, the resistive survey within the walled garden, in 2002, showed no low
resistance in the garden on that line. White's Cramond plan of 1797 (Ref. 8.11) calls
this feature a "sunk fence" and it is apparently only McGlynn who refers in Ref. 8.2 to
"a grass walk, presumably along the top of the sunk fence or ha-ha". The section
appears unlike a ha-ha which would be expected to slope away gently on the N side.
The term "sunk fence" could imply a fence placed at the bottom of the ditch; this
would perform the same fimction as a ha-ha in excluding cattle from the garden or
park area whilst making the fence invisible from the house. Holmes (Ref. 8.10)
comments that "to the south of the ditch (Area VII) was a cobbled footpath. The
surface of the footpath sealed pottery of 17" century type". If the Area VI1 ditch does
relate to the "sunk fence" and to the linear low resistance along the raised beach, the
implication is that the ditch existed, and was utilised, as a garden boundary probably
at the time of building of the 1680 house. The cobbled path could have been made at
that date and been bottomed with the pottery. It is significant that these features
continued to be used and recorded by Bauchop in 1815.

The linear array measurement 8 (Illus.7.5)was made in an easterly direction at the


junction of grid squares G30 and 3 1 and extended 4.0m, hrther to the E. This crossed
the less clear low resistance where the ditch appears to curve to the S on the raised
beach edge adjacent to the burn. At this point on the linear array printout, between the
12.0 and 14.0m. marks, the low resistance rises to a depth that would just be
detectable in the area survey but does not give a clear sectioned ditch picture. The
ground here, because it is near the burn, is wetter and of lower resistance and may
have been altered by slumping.
The 120m, linear array line 4 (Illus.7.5), made with electrode spacing of 5.0m.,
extends 40m. N of the area survey and at its extremity shows a high resistance that is .
probably due to a pipeline and its bedding stonework. Between the 56 and 80m. points
it is co-linear with line 3 but lacks the detailed resolution of that line due to the
electrode spacing being increased by a factor of five. The large high resistance
between the 40 and 70m. marks is likely to be a gravel deposit in rear of the raised
beach; the E-W length of this deposit can be seen in the area resistance survey.
Shallower high resistances at the 7, 17 and 33m. marks align reasonably with small
higher resistances on the area survey and could be either archaeological or of
quaternary geological origin.
The two magnetometry surveys (Illus. 7.6 and 7.7) indicate a number of anomalies
that align approximately with resistive features shown on Illus. 7.3 and interpreted in
Illus.7.4.This interpretation has mainly indicated low resistances whereas the linear
magnetic anomaly that runs along the N edge of squares G17,21 and 25 appears to
align better with the high resistance on the N side of the ditch. Similarly the magnetic
anomaly that is on the S side of squares G16, 20, 24 and possibly 30, has interpreted
resistive lines'running on either side of it. These, together with the clear rectangular
feature in square G23 (Ill~s.7.3)~ all align with the tree avenue and point strongly to a
garden layout that predates the conversion to parkland. As suggested by McGlynn
(Ref.8.2) this garden was probably contemporary with the 1680 house.

The interpretation of the area resistive features on the "Roman" alignment in squares
G 1 1,14 and 15 closely parallels the interpreted shapes based on the magnetometry
survey (Illus.7.6) as does the NE-SW section of the magnetic anomaly in square G12.
The magnetic survey, with readings taken at 1.Om, spacing, over squares G 18,19,22
and 23, Illus. 7.7, shows more detail than that in Illus. 7.6. The magnetic anomaly that
runs at an angle across square G 18 is more clearly linear and a rectangular outline in
square G 22, although less clear, is discernable in Illus. 7.7 but not on that line in
Illus 7.6. The interpretation of the resistance plot over these two squares indicates the
rectangular outlines of what could possibly be two buildings on the Roman alignment.
The two magnetic anomalies in these squares in Illus. 7.7 are in the same area as the
resistive anomalies but do not appear to be on the Roman alignment. The re-
interpretation of the magnetic anomaly shape in G22, Illus. 7.6, could be seen as
bringing the more easterly of these magnetic anomalies into coincidence with a
resistive interpreted feature in Illus. 7.4.

There is sufficient correlation between resistive and magnetic anomalies to suggest


that features exist on the Roman alignment and represent paved areas or wall
foundations although some, that are low resistance, may represent robbed out walls or
. dirt paths between buildings. The misalignment of magnetic and resistive anomalies
in squares G 18 and 22 may be due to incorrect interpretation but, as medieval
building was found within the Cramond bath house, there could have been medieval
structures also built in this area.

The interpretation of geophysical data is far from an exact science but the correlation
of the anomalies from these two surveys and from the particular data derived from the
linear array measurements, gives confidence that there are both garden archaeology
from around 1680 and Roman remains below the parkland. Other period remains
could also exist where features are on neither the garden nor the Roman alignments.
The estate plans of White and Bauchop add information on the line of the "sunk
fence" which appears to give fbrther endorsement to this being of Roman origin albeit .
not on the line of other Roman features. If the formal garden was converted to park
land about the date that the E wing was added to the house in the 1770's the ditch was
obviously retained to be recorded by Bauchop in 1815.

These conclusions are, of necessity, tentative as they are limited by the fact that the
investigation, although employing three separate geophysical techniques, was non
intrusive. Confirmation of the main thrust of the conclusions, that there are both
garden and Roman features beneath the parkland, could probably be confirmed by the
excavation of three strategically placed trenches.
6. Acknowledeements

We are pleased to record our thanks to:-

Historic Scotland for permission to survey over the parkland much of which is within
the Scheduled Monument area. Further thanks are due to Historic Scotland for
providing a grant towards administration, travel, expendable equipment, insurance
and the preparation and publishing of this report.

Davidson and Robertson (Mr Niall Milner) Land Agents for Cramond and Harthill
Estates, the owners of the parkland, for permission to access the area.

Dr. Bruce Hobbs, Dr. Roger Scrutton and MI-Greg Macdonald of the Geosciences
Dept. of the University of Edinburgh for their cooperation during the making of the
resistive linear array and magnetometry surveys. Special thanks are due to Mr Greg
Macdonald for providing a copy of his report with permission to reproduce his
printouts.

Mr Mike Middleton of Headland Archaeology Ltd. for providing an interpretation of


the garden and Roman features on the area resistive survey.

The Society members involved in the layout, area resistance survey and topographical
readings were :-
Kyle Armstrong, Alan Calder, Ron Cameron, Patrick Cave-Brown, Kathryn Cox,
Steven Cox, Valerie Dean, Amanda Dickson, Carol Dickson, Ian Hawkins, David
Jones, Bob Marks, Don Matthews, Bill McLennan, Inez Mepham, David Metcalf,
Tom Sharp, Denis Smith, Jill Strobridge, Irene Taylor and Peter Tothill.

Further thanks are due to Val Dean for representing the Society on the Cramond
Management Group during the preparation period of the Marta McGlynn Associates
Management and Interpretation Plan and for other inputs vital to this report.

You might also like