Centuriation? on the Hoo Peninsula
Introduction
In Archaeologia Cantiana Volume Lxv ( 1952) Michael Nightingale wrote an article entitled, A Roman Land Settlement near Rochester (p. 150-159). His article describes and discusses the possibility that part of the road and field system to the north of Rochester, on the Hoo peninsula, is divided up into a Roman period ce11turinl grid based upon the nc/11s. I am the first to admit that 1 have never taken much notice of the article for land division, of any period, is not something that interests me as an individual. However, in 1996 local Medway archaeologists and historians approached the present writer with an article about work undertaken on the Isle of Wight by Mr. C. T. Witherby. ln this article he mentions the work of Michael Nightingale and I was asked to comment on the 1952 article in an attempt to help Mr. Witherby. For far too long I ignored this request for the simple reason I knew that the comments I could make, would entail considerable work. In October 1998 the f"'-urator of the Guildhall Museum, Rochester (Michael 1v1oad) also approached the present writer with a similar request, this time on behalf of the Mayor of Cesena, Italy. The authorities of Cesena are undertaking a worthwhile European wide study on Roman centuriation of agricultural land. For someone who is mathematically disadvantaged the amount of mental pain suffered whilst undertaking this work has, at times, bordered on the unbearable. I assure readers this article was more painful for me to write than for you to read.
Centuriation
Roman centuriation or the regular grid like laying out of fields, of which superb examples exist in Italy, is based upon a series of large rectangles, those at Cesena measuring 707 by 710 metres and at Hoo a square of 710 metres was suggested (Nightingale 1952, p.151). The term centuria refers to one of these squares which can be divided into 100 (hence centuriation) units or heredia. Each of these smaller units can be divided into two iujera, which correspond to the amount of land a team of oxen could work in a day. The i11gen1111 measure 2 by 1 nc/11s, and the sides of an nctus measure 120 Roman feet or 35.48m. (one Roman foot = 11.6496 English inches or 29.57cm).
Some of the roads, tracks and field boundaries on the Hoo peninsula seem to correspond to an nch1s or a multiple thereof. On page 156 of his article Nightingale states that many of the fields in this area seem to be based upon a length of 5 nctus rather than an English furlong. This may well be so, but it is very noticeable that the aspect of English measurements is not dwelt upon and that multiples or divisions of furlongs are ignored. Michael Nightingale's article is a superb example of deductive interpretation but, the failure to compare and contrast with alternative types of land measurements weakens his case considerably. One of these measurements, which historians and archaeologists ignore at their peril, is the English rod, pole or perch. When I first came into archaeology if anyone had told me that one day I would write about the rod, pole or perch I would have regarded them as a 'nutter'.
A refresher course on imperial measures
1 mile = 1760 yards= 8 furlongs = 5280 ft. (English).
1 furlong = 10 chains = 220 yards and also 40 poles.
1 chain = 22 yards = 4 poles.
1 pole = 5.5 yards or 16.5 feet.
The term rod, pole and perch are all the same measurement. The word pole is used throughout this text and was used as a measure of land certainly in the postmedieval period and presumably at an earlier date. The pole is apparently of great antiquity, being equated to Sumerian cubits (Everyman's Encyclopedia).
The supposed centuriation on the Hoo peninsula
The reader is referred to Michael Nightingale's article for a detailed description and discussion of the possible ce11t11rinl land division to the north of Rochester. It is obvious to all who study an Ordnance Survey map (the 1:25000 or larger scale are the best to use) that some of the roads and consequently the fields in the area to the south of Cliffe and Cooling run (more or less) parallel to one another. The north to south parallel axis of the roads is more noticeable than any parallel east-west axis.
There are six main north to south aligned roads or tracks, starting on the west side of the map (Fig.I):
a. West Street
b. Station Road.
c. Well Penn Road and its northern extension Thatcher's Lane.
d. Rye Street/ Cooling Street.
e. A road extending (south to north) from Spendiff Farm to Alma House.
f. A road extending (north to south) from Cooling Castle to Cooling Court and New Farm.
In his work Nightingale regards road 'd' as the principal axis or Cardo Maximus. West Street (road 'a') is a further cardo and road 'c' is at a distance of 10 actus west from the Cardo Maximus and 'b' 30 actus to the west. He makes no mention of road 'e' or 'f'. After a considerable swing to the west the northern part of Thatcher's Lane seems to be on the 20 actus line, but it is not mentioned by Nightingale.
The Cardo Maximus
Road 'd' has the longest straight section of any of the roads in the area and was chosen by Nightingale to be this principle axis. By itself this choice would be purely arbitrary. However, Nightingale argued that if this line is projected it lines up with the Rochester to Maidstone Roman Road to the south of the river Medway. This latter road would then have formed the surveying line for the centurial land system to the north of the river. To obtain a regular grid a further line was necessary at right angles to the first and this was supplied by extending Watling Street in a straight line towards the River Medway. This axis formed the Decumanus Maximus.
One of the great fallacies of our time is that Roman roads are constructed in straight lines. They are constructed m straight sections, but if they just went in a series of straight lines we would have a completely different settlement pattern of Roman towns than the one that actually ex�sts. It appears that Nightingale had taken the !onge _ st stra1�ht length of road in the centurial and then lined 1t up ":'1th one of the sections of Roman road to the south of the nver and then equated this with a surveyed line (A on Fig. 1). The longest length of Roman road to survive to the south of the river is that which passes through Delce. If projected this gives a line several hundred metres to the west of the longest straight length of the Cardo Max1111us (Bon Fig. 1). It also more or less aligns with road 'c'. Why then, isn't 'c' the principal axis? Which part of the Rochester to Maidstone road should be used as the survey line? If another section of Roman road �s taken (C on Fig. 1) and projected, then an even gre� ter d1scr _ epancy to the east is arrived at. If any of the straight sections (at least five) of the supposed Cardo Maxim us to the north of the river are used not a single one of them mat0es up with the Delce section of the Roman road. Usmg the Roman road to the south of the river, as a survey line seems to be meaningless and therefore, likewise creating a right angle from it to form the Decumanus Maxim us. The 2 evidence for the existence of a cenlurinl system to the north of the river therefore has to depend upon what actually exists on the Hoo peninsula.
The centurial roads and fields
The roads to the south of Cliffe and Cooling whilst parallel are only so in a general sense. To use Nightingale's terminology they are 'warped' (p.157) i.e. there has been a shift in the alignment of the road. Through a process of time, warping could easily take place, although the shift in roads 'a' and 'b' seems excessive. Those who would argue centuriation exists, seem to have taken the specific points on roads 'a' to 'd' which can be fitted into an actus system. The idea of warping or shift whilst perfectly possible, indeed probable, creates an obvious problem. There is no way of knowing whether the points used on any of the roads from which measurements have been taken are actually in situ Roman roads. The genuine Roman road (if any) may be the part that appears to have shifted. We have no way of knowing whether or not a specific section of road has moved, nor do we know the date of the original construction or movement (if any).
Taking one example chosen at random, Station Road. (road 'b') is 30 actus to the west of the Cardo Maxnnus along line 1. However, if the measurement is taken further south along line 2 then the measurement in actus is approximately 26.375 actus which doesn't make a lot of sense. However, in imperial land measurement:
26.375 x 120 Roman feet x 11.6496 inches divided by 12 = 3072.582 (English) feet, divided by 3 = 1024.194 yards, divided by 5.5 = 186.22 poles= 4 furlongs 26 poles 3ft and 5 inches.
For the sake of argument this could be regarded as 4 furlongs and 26 poles. I have no way of knowing whether it was normal to measure in single poles. However, I have no doubt that measurements in 4 pole (i.e. a chain) multiples were made. Also I have no doubt that by taking specific straight stretches of road, distances in 4 pole multiples could be created over much of the area.
The length of the road to the south of line 2 is on • different angle, but just as straight as that at 1. To take a length of road and say that it is 30 actus away (even thought the rest of the road is not at that measurement) from (an unproven) Cardo Maximus and that therefore it existed in the Roman period seems to be the height of archaeological folly. 'Yes', more or less exact act us measurements can be created. However, it would be just as logical to take a stretch of road on a 4 pole multiple to the 'Cardo Maximus' and say that the road at that point was measured out using English land measurement in the post-medieval, medieval or Anglo-Saxon periods. There is just no way of knowing.
The same applies to the fields themselves. A measurement of 5 actus seems to apply for some fields, but some fields seem equally to have been laid out using furlongs and also:
5 actus = 600 Roman feet = 582.48 English feet (600 x 11.6496 divided by 12),. divided by 3 = 194.16 yards = 0.971 furlongs= 35.3 poles.
If the measurement we�e measured with a tape over this 582.48 feet (remembering the land is not flat) a discrepnncy of 0.3 of a pole (4.95ft) from a measurement of 35 poles i not bad. In my view the most likely method of measuring out a field boundary would be by pacing or u ing a 'pole' of known length. If so the measurement then becomes reasonably (one is tempted to say 'very') accurate. Therefore the 5 act11s fields are just as likely to be 35 poles. If a figure of 36 poles was used (i.e. 9 chains) even then the error if placed over undulating, and no doubt muddy land, is not excessive (about llft).
Furthermore although the northern part Thatcher's Lane seems to be within Nightingale's limits for actus, it is, along with road 'e' and 'f' ignored. The latter two roads do not seem to match up with actus measurements as taken from the supposed Cardo Maximus even though they could (arguably) be regarded as parallel to the others. The reader is referred to Nightingale's Figure 2 (based on the O.S. 6 inch map) for the following point. On this figure it is noticeable that if projected northwards the line of road 'c' does not line up with field boundaries as taken from the 1840 Tithe Map. This projected line cuts through the fields, whjch are supposedly of Roman date.
The longest alignments of the field boundaries as taken rfrom the Tithe Map are aligned both north to south and •east to west. Others are angled away from these main ruspositions; overall the fields are not in alignment with one another and therefore can hardly be used to argue in favour of a centurial system.
Maps
Over the years I have learnt that trying to use 1:50000; 1 :25000; 6 inch or other 'normal' scales for thjs type of work can only create an impression of what is there. To obtain accurate measures for field boundaries, distances between roads, etc. the Ordnance Survey 1:500 scale maps need to be used.
In this instance there is also the problem of tracing the field boundaries from the 1840 Tithe Map, then having to convert them to the O.S. 6 inch scale, and then actuaJly drawing them in their correct position. Such a process is difficult, errors creep in. In Nightingale's Figure 2 for r"1"1stance the scale of furlongs itself has been incorrectly Jrawn; 1.7cm. represents 1 furlong, yet 2 furlongs are 3.55cm. (should be 3.4cm) and 3 furlongs 5.4cm. (should be 5.lcm.). The discrepancies are small but once scaled up they are enough to create considerable errors. For example comparing the discrepancy between 5.1cm. and 5.4cm once scaled is 13.9 yards•= 41.7ft. which is just over 2.5 poles. Throughout I have taken the 1.7cm. measurement (and multiples thereof) to be representative of furlongs.
Conclusions
It seems that in Nightingale's article we have a classic case of something that can be seen on maps to be unusual (a point upon which everyone agrees) and then archaeologists allowing their own biases to overcome objective thought. I would suggest that neither the roads nor the fields can satisfactorily be used be used to argue that a Roman centurial system was laid out in this area. Specific points can be taken and the measurements can be made into whatever the researcher wants. We all know (but may not admit) that in such instances there will be a 3 tendency to allow our own biases to dictate our interpretations. I have no doubt that if we really tried a case could be made for tenth century B.C. Sumerian colorusts measuring out the land in cubits.
As I have been in archaeology a long time I am not so stupid as to say that a Roman centuria/ land system does not exist to the south of Cliffe and Cooling. It may do, but the case for such has not been proved beyond reasonable doubt. Equally valid alternatives based on the furlong, chain and/ or pole could be created.
Those individuals who believe that such a Roman centurial system exists in this area and who want to convince sceptics such as myself should, in my view, first undertake a study of estate maps. I suspect that the land formed part of the Cobham Hall estate for wh.ich maps exist. Such a study may give an earlier date (if very lucky perhaps back to the early seventeenth century) for the field system shown on the 1840 Tithe Map. Then by nonexcavation archaeological techniques (overall aerial photographs and geo-physical surveys in selected places) a planned field system may be identified in the ground.Excavation in selective places to obtain dating evidence from any ditches that might survive could then take place. By plotting out this evidence on the largest scale map they would then be able to show whether a Roman centurial grid was created in this area and I would be the first to congratulate them.
NOTE: I have been told the Uruversity of East Anglia have undertaken a computer study of the 'centuriated' area around Cliffe and have proved that it exists. The obvious first question to ask is has a comparative computer study been undertaken? Have computer grids using (surprise, surprise) the English foot, pole, chain or furlong-as well as the Roman actus been created? I have the suspicion that no such comparative study has been made and until such time that it is undertaken any 'proof' that a Roman centuriated system exists is worthless.
Addendum
To the south of the River Medway, along the length of Watling Street from the eastern outskirts of Gillingham as far as Newington, there is a similar parallel road system. This apparent regularity has itself been commented upon (most recently in the superb new book The Gillingham Chronicles by Ron Baldwin 1998. Many years ago when I attended Ron's adult education classes in local history he always told us to go out and 'do it'; observe, listen, collect the evidence, write it down. At the time I never thought I would. Now I do and, most of the time, I thoroughly enjoy the chaJJenge. Thanks Ron).
Several of the roads Barnsole Road, Woodlands Road, Featherby Road, Eastcourt Lane and Twydall Lane are all known to date back at least to the fifteenth century and some at least may have been tracks in the Roman period (Baldwin 1998, p.32-3). However, if the reader studies the Ordnance Survey 1:25000 map is very noticeable that this western group 'kinks' about halfway between Watling Street and the river and creates a djfferent parallel alignment. When areas to the east are sturued two further groups of parallel roads also appear. It is very noticeable that only with the easternmost group of roads, nearer to Newington, is anything like a right angle created with Watling Street. It is also noticeable that this group is the closest to Hartlip and Bo ted Roman vi1las, which may (or may not) be significant. The western group of roads are nearer a right angle with the Lower Road adjacent to the river and one is tempted to say they were offset from that road rather than Watling Street and the latter formed the southern termination of these roads.
Whilst groups of roads are parallel to one another the groups are not. If this is a Roman land system it can hardly be said to be regular let alone ce11fllrial. I have only measured the distances between the eastern most group of roads. Along the random line chosen neither in actus or furlongs did the measurements seem to make any pattern. Whilst the 1:25000 scale is unsatisfactory for such a test I suspect that the general conclusions would not be altered using a larger scale map. As with the area to the north of the river if selected points were to be chosen I am quite sure we could create patterns in whatever measurement we want. Such a way of making a field pattern, of any period, will be meaningless.
NOTE: A more detailed text, along with several figures, has been deposited with the Guildhall Museum, Rochester.
Alan Ward
May 1999
Figure 1. The 'parallel' road system in the Rochester area.
