MELANDRA ROMAN FORT, GREATER MANCHESTER - SELECTED CERAMIC BUILDING MATERIALS

S Garside-Neville

The tiles described are a selection of material from the 1973, 1974 and 1977 seasons of excavations directed by J P Wild of Manchester University.

FORMS

CAVITY WALLING

There are two forms of ceramic building material specifically connected with the Roman hypocaust heating system in the Melandra sample. These are half box flue and box flue.

Half Box

This form of tile (Fig. 1) is currently thought of as being the predecessor of the more familiar box flue tile. It was used up against the wall to create a cavity walling for the circulation of hot air. Cutouts allowed the movement of air laterally along the wall. Shaped like a tegula, with flanges along two opposing sides, it is relatively easy to distinguish from the roof tile. The flanges tend to be larger, with cutouts in a section approximately in the centre of the flange, and a sanded underside scored with a diamond pattern, or occasionally combed (Brodribb, 1987, 67). The patterning serves as keying for mortar and plaster, and in the Melandra examples is formed by drawing a sharp knife or stick along the surface. There can sometimes be a 'signature' on the upper surface, which may have indicated the identity of an individual tile-maker, though a signature has not been observed on the Melandra examples. The Melandra half box tile measurements (Table 1) seem to be smaller than the average (Brodribb, 1987, 143) except for the flanges, which are up to 6mm taller than the average.

_________________________________________________________________

Table 1 - Measurements of the Melandra half box tile

(Measurements in millimetres)

Context Fabric Length Breadth Thickness Flange Height

u/s F1 ? 310 28 85

VI (2) F1 ? ? 35 86

" F1 ? ? 28 80

" F2 ? ? 31 80

" F2 ? ? 30 ?

" F2 ? ? ? ?

" F1 350 ? 35 85

" F1 ? ? 32 ?

" F1 ? ? 29 ?

" F1 ? ? 32 ?

" F2 ? ? 32 86

" F2 ? ? ? 78

" F2 ? ? 34 ?

" F2 ? ? 31 ?

________________________________________________________________

Box flue

An improvement on the half box tile, the square pipe-shape of the box flue was easier to use, being less cumbersome. The Melandra examples, where complete enough, are quite square in section (Fig. 2), whereas the norm appears to be a more oblong section. The one complete example of box flue matches the average length (Brodribb, 1987, 143), but because it is very square it does not match the average width (see Table 2 for measurements). There are two types of keying on the Melandra examples. There is knife scoring as used on the half box (see Fig. 1), and there is also scoring with a fairly blunt object (as Fig. 2). The illustrated example clearly shows how the box flue was formed, which was by wrapping a flat length of clay around a mould and making the overlap join in the centre of a flat side, which confirms the experiments carried out by Morgan (1979, 395). The fragments of box tile examined were sooted on the inside, which may indicate that they were close to the furnace.

_________________________________________________________________

Table 2 - Measurements of the Melandra box flue tile

(Measurements in millimetres)

Context Fabric Length Breadth Thickness Keying

u/s F2 366 145 18 Scored

VI (24) F2 350 136 22 Knife

" F2 ? 120 20 "

" F2 ? 120 21 "

" F2 ? 110 19 "

" F2 ? ? ? "

" F2 ? ? 18 "

" F2 ? ? 21 "

_________________________________________________________________

BRICK

Brick is defined as being ceramic building material used in the walls and hypocaust floors. In the Melandra samples there were miscellaneous fragments that could have come from brick, but the size of the brick could not be determined. Other larger fragments could be assigned specific forms, and therefore more specific functions.

Cuneatus voussoir

These wedge shaped tiles were used to form arches. Unfortunately there are no complete examples, although the length of some of the fragments points to an oblong shape, but the tapering to form the wedge is observable as it occurs along the breadth of the tile (Table 3). These tiles conform roughly to the average (Brodribb, 1987, 142).

_________________________________________________________________

Table 3 - Measurements of the Melandra cuneatus voussoir

Context Fabric Length Breadth Thickness Thickness

1 2

VI(3) F1 ? 230 63 38

" F1 ? ? 63* 35

" F1 ? ? 58 43*

" F1 ? ? 50 40*

VI(24) F1 370* 225 60 39

* Broken edge, minimum thickness/length

_________________________________________________________________

ROOFING TILE

The Roman roofing system consisted of tegulae (tile with flanges) and imbrice (curved tiles that covered the joins made by two tegulae with the flanges side to side).

Imbrex

There are two complete examples of imbrices from Melandra. Both are overfired, and have suffered some distortion. The tiles both have a finger smoothing mark around the end of the tile, and have been tidied up just after moulding by running a finger or knife along the edges (Fig. 3).

_________________________________________________________________

Table 4 - Measurements of Melandra Imbrices

Context Fabric Length Breadth Thickness

(widest

part)

VI(27) F1 455 155 18

" F1 450 170 19

_________________________________________________________________

Tegula

There are only small fragments of tegula from the sample. However, the examples present do have some interesting features. The fragment from X(1) was overfired, had a signature at the end of the tile (in the form of an arc, made by two fingers), and also had a lower cutaway (to help join the tegula to the next tile) which indicates that it came from the bottom of the roof tile. A piece from XIV(110) also had a lower cutaway and had a possible tally mark on the back of the flange. The mark is in the form of a 'V' and may indicate the number of tiles in a batch made on a certain day, or was perhaps a personal identifier for the tilemaker (Brodribb, 1987, 135).

_________________________________________________________________

Table 5 - Measurements of Melandra Tegulae

Context Fabric Length Breadth Thickness Flange height

X(1) F1 ? ? 28 ?

XIV(110) F2 ? ? 34 69

_________________________________________________________________

FABRIC

Four fabrics were observed and have been described using a x10 hand lens:

F1 Red-brown in colour, often overfired to deep purple. Abundant quartz up to 1mm, occasional white quartzite up to 3mm.

F2 Brown in colour. Occasional pebbles, abundant quartz up to 1mm, occasional ?black iron ore, occasional clay pellets

F3 Brown in colour. Abundant quartz up to 1mm, occasional straw or grass, occasional silt, occasional pebbles up to 2mm, occasional clay pellets.

F4 Brown, fine quartz around 0.5mm, occasional quartz up to 1mm, occasional mica, ?limestone up to 2mm.

Fabrics F1 and F2 are very similar in appearance. These two fabrics comprise the major part of the sample, with products including box flue, half box flue, cuneatus voussoir, imbrex and tegula. The other two fabrics, F3 and F4, are confined to unidentifiable forms.

CONCLUSION

Although only a small sample, it is possible to illuminate the method used for building the cavity walling for the hypocaust system. The two different forms - half box and box flue - may hint at rebuilding as the two formats are not necessarily compatible. For the half box there is evidence from elsewhere that these tiles possibly went out of use in the late first or early second century when box flue tiles are thought to have become popular (Betts, 1985, 151) though there may be later examples from a few sites (Brodribb, 1987, 67; Crowley, 1995, 150). The blunt object scoring of the box flue echoes the combed keying found on box flue tiles which is thought to be a later phenomenon, and which would have been a much more efficient key for mortar and plaster than the narrow slashes made by the knife scoring (Brodribb, 1987, 109).

The other forms present (roofing and brick) indicate only a general trend, and it would be fruitful if a more detailed study of the material recovered from the site could take place.

With at least four fabrics present on the site, it may be an indication of different kiln sites supplying the fort. This may have been over a long period of time, which again would help pinpoint different phases in the fort's occupation. This, along with study of the variety of forms present, would characterise the products found on the site, illuminate further building phases, and help set the material in its context within the brick and tile industry of the area.

ACKNOWLEDGEMENTS

John Peter Wild for allowing me to report on the tile. Kurt Hunter-Mann for his drawings.

REFERENCES

Betts, I M, 1985 A scientific investigation of the brick and tile industry of York to the mid-eighteenth century. University of Bradford, unpublished Ph.D

Brodribb, G, 1987 Roman Brick and Tile. Alan Sutton

Crowley, N, 1995 'Building Material', IN Cowan, C 'Possible mansio in Roman Southwark: Excavations at 15-23 Southwark Street, 1980-86', Transactions of London and Middlesex Archaeological Society, Volume 43, 1992, 144-164

Morgan, G, 1979 'Experiments in making and firing box-flue tiles', IN McWhirr, A (ed), Roman Brick and Tile: studies in manufacture, distribution and use in the Western empire, BAR International Series 68

S Garside-Neville

Brick and Tile Services

August 1995

RECORDING SHEETS

SITE CODE: MEL 74

MELANDRA: CERAMIC BUILDING MATERIALS RECORDING SHEET

Context: VI (2) Period Phase

Earliest Date Latest Date

Fabric No	Form Name	Corners	Weight (gm)	Length (mm)	Breadth (mm)	Th mm	Mortar	Comments 1. Kept 2. Discarded
F1	Roman misc	1	825	-	-	46	-	1. Possibly a cuneatus
F4	Roman misc	2	3100	232	-	48	-	1.
F1	Half box		2000	-	-	35	-	Flg ht 86mm, cutout (140mm), diamond knife scored,1
F1	Roman misc	2	4000	273	-	62	-	Thick at edges, in middle 47mm, 1.
F1	Roman misc	1	825	-	-	41	-	Finger prints on edge, 1.
F1	Roman misc	-	375	-	-	-	-	2 fragments, 1.
F1	Half box	1	1300	-	-	28	-	Cutout, diamond knife scored, flange ht 80mm, 1.
F2	Half box	-	1500	-	-