Kent Archaeological Review extract

'The Painted House' Underfloor Heating System.
by John Willson.

In Dover we have perhaps the best preserved example of a Roman under- floor heating system yet found in Britain.

This system was discovered during rescue excavations carried out by the Kent Archaeological Rescue Unit in Dover's town centre in 1971. It reflects the method used to heat at least six rooms of a Roman house, better known as the famous 'Painted House'. As has been previously explained, this rather sumptuous system of underfloor heating was common to villas and town houses in Britain owing to the colder climate (in Mediterranean countries braziers and fire-places had a more general distribution). It is therefore no surprise to find that such a sophisticated building had such well-planned heating arrangements.

There are two types of hypocaust system common to Britain and these can be usefully described as the 'channelled' and 'pillared' types. By combining these two types we achieve a 'composite' system where the main flues are 'channelled' and open into a small 'pillared' area in the centre of the room. This is exactly what has been found in Dover.

GRAPHIC: Figure 1. Plan of the Painted House, Dover 1971.

Figure 1. Plan of the Painted House, Dover 1971.

Rooms 1 and 2 showing underfloor heating system.

Scale 1 :100 cms.

  • 1. Later Fort Wall.
  • 2. Central Chamber.
  • 3. Flue Channels.
  • 4. Mortar Floor.
  • 5. Furnace Arches.
  • 6. Wall Flues.

An examination of the heating arrangements of the 'Painted House' shows that each room was fed from an external furnace through a furnace arch (Figure 2). Excavation has not revealed any stone constructions surrounding the stoking areas and it would appear that the fire was contained within a lean- to placed before each individual furnace arch. The hot air and gases from the fire passed through the furnace arch, along the main flue channel to the central 'Pillared' chamber and then circulated through the other seven flue channels which radiate from the central chamber in 'Union Jack' fashion (Figure 1) to the sides and corners of the room. Each of these channels terminates in a vertical wall flue composed of box flue tiles built into the fabric of the wall. These vertical flues presumably conducted the waste gases to ceiling or roof level and expelled them into the air. For readers requiring a more detailed examination of the various components of the system, each aspect can be dealt with separately.

GRAPHIC: Figure3. Section through a typical hypocaust channel.

Figure3. Section through a typical hypocaust channel.


This furnace arch, (one of three actually found) was constructed of red tiles set in mortar and built into an external wall adjacent to the furnace area. The arch consists of some 14 tiles, 42 by 30 by 4 centimetres set in the form of a semi-circle capping a vertical sided channel constructed of 12 tiles set in mortar, giving an overall height of 60 centimetres and a width of 45 centimetres. Tiles were used in the construction of the furnace arch, main feeder channel and the pillars in the central chamber as they may have withstood the great heat produced from the furnace far better than natural stone.

GRAPHIC: Figure 2. South Wall of Painted House.

Figure 2. South Wall of Painted House.


The flues of the channelled hypocaust are cut into the surrounding soil and are on average 65 centimetres deep by 30 centimetres wide. The sides of the channels are built of chalk blocks set in chalk wash, spanned by two to three tiles 42 by 30 by 4 centimetres which in turn are sealed by a large capping tile 42 by 30 by 4 centimetres. The whole construction is bonded with orange clay, upon which a flint and chalk hardcore and finally the mortar floor (opus signinum) some 13 centimetres thick, of the room was laid. The floors of the flue channels are often of soil or mortar but in this case, however, the mortared floor of an earlier building is used as a base.


The pillared area lies in the centre of the room in the form of a chamber approximately 1.75 metres square by 70 centimetres deep. The pillars were built firstly to support the great weight of the floor in this central area where all the flue channels meet and secondly to give an area where the hot gases could circulate and disperse to the radiating channels. The pillars (pilae) were constructed of some 13 main tiles generally 24 x 20 x 3 centimetres laid one on top of another and bonded with clay to a height of 53 centimetres. The pillars rested upon the mortared floor of an earlier building and were capped with two tiles of a larger size 28 by 28 by 4 centimetres. The pillars were spaced at approximately 60 centimetres. (from centre to centre) to give the required support to the floor.

GRAPHIC: Figure 4. Diagram showing the hot air flow in a 'composite' hypocaust system. 
This arrangement is similar to that found in the 'Painted House'.

Figure 4. Diagram showing the hot air flow in a 'composite' hypocaust system. This arrangement is similar to that found in the 'Painted House'.

THE WALL FLUES (Figures 1 & 4).

The wall flues were placed at intervals around the room to coincide with the heating channels below the floor and were basically purpose made special tiles commonly known as 'box flue' tiles. They were generally oblong in shape and open at both ends, approximately 2 centimetres thick, 36 centimetres long, 19 centimetres wide across the face and 13 centimetres, across the sides.

The sides of these box flue tiles were usually scored so as to form a key for the adhesion of the mortar when they were laid in position. The scoring on the box flue tiles comes in many varieties and designs ranging from simple combed crosses to latticed patterns, 'Union Jack' designs, curved and wavy line decoration all scored into the surface of the tile before it was fired in the kiln. The lateral openings often seen in these box flue tiles were to allow the hot gases to pass from one to another freely to equalise the heat, of course this was only possible when two or more box flue channels were in tandem.

Having thus examined the plan and purpose of the system it is only possible to estimate its effectiveness based on the account given by Mr Keller. None the less it is not difficult to imagine how comfortable it must have been to live in a house such as this provided with numerous heated rooms where not only the floors were pleasantly warm, but the walls were acting as additional radiators. Added to this of course the house was decorated with superb wall paintings and a fine pink mortar floor (opus signinum). For residents enjoying such comforts it must have felt much like being in the heart of Rome. The lot of the stoker, however, was perhaps a far less envious one. In cold weather he would have the prolonged task of firstly, the initial firing until the thick floors absorbed enough heat to retain a constant temperature and secondly, being prepared to stoke all of the individual fires about twice a day to keep the interior pleasantly warm. Presumably the cost of obtaining enough fuel to see the winter through must have been an important factor, but was certainly justified by the comfortable living conditions obtained. Nor did the slaves, used to service the system, threaten a total shut-down in mid-February for their own dubious ends !

Close Window
Accessed this page via search engine or bookmark? Full K A R Index here.