Explaining the bookend effect
It has long been recognised that flank walls of Georgian and Victorian end-of-terrace properties are prone to bulging or leaning outwards. In most cases the cause is local to the flank wall, such as inadequate connections to internal walls and floors, sulphate attack of the masonry, eccentric bearing of floors and roofs, high winds or even residual bomb damage from the Second World War.
However, sometimes the cause is deep-seated within the terrace, due to the lesser-known 'bookend effect', whereby longitudinal expansion of the terrace thrusts the end-of-terrace properties sideways as a whole, including their flank walls.
Symptoms of distress
Not only are the flank walls pushed out of plumb, but the front and rear walls are racked sideways in their own plane, becoming out of square. The racking is manifest in the lozenging of door and window openings, and most pronounced in the end-of-terrace properties. Internally, spine walls are similarly distorted. Party walls lean in the same direction as the flank walls. The net effect is wholesale lateral distortion, like a failed bookend.
Doors and sash windows may stick; gaps may open between floors and party- walls. Sometimes vertical cracks develop in the brick aprons beneath windows. Brick arches and stucco hoods above windows tend to drop.
Where terrace properties, such as shops or public houses, were built or converted with large wall openings at ground-floor level, the lateral distortion is concentrated in the ground storey where sway-stiffness is least. The superstructure above remains relatively plumb.
Longitudinal expansion of continuous front and rear terrace walls will displace sideways the party and flank walls connected to them. Expansion is greater at the top of the outer walls due to their greater exposure, lesser heat-sink effect and less pre-compression. As a result the crosswalls are displaced more at their tops and so lean sideways (see diagram).
When the front and rear walls subsequently contract, they have insufficient tensile strength to pull the displaced crosswalls completely upright again. So the front and rear walls either crack vertically at their minimum section - on the line of door and window openings - or the soft lime mortar allows individual bricks to draw apart from each other by minute amounts. Debris and dirt find their way into the cracks or gaps between bricks, preventing their complete closure during the next expansion. Overall, the front and rear walls expand slightly each cycle - the 'ratchet' action of crack growth. In theory, the centre of the terrace is stable, with distortion progressively increasing towards the ends-of-terrace.
The average Georgian or Victorian terraced building in question is between three and five storeys high, its openings facing front and rear, with a central spine wall. External and party walls are solid brickwork, usually set in lime mortar. Spine walls are typically solid brickwork at lower levels, and timber studwork higher up. Partitions are usually timber studwork. Suspended floors and roof carcassing are of simple timber construction. (Back additions are ignored here.) Almost every wall and partition usually assists in delivering vertical building loads to the foundations. Lateral stability is conferred principally by the in-plane shear resistance of the walls.
The following four London case-studies illustrate the bookend effect, its symptoms and remedial measures.
A Georgian shop-fronted terrace
No 117-123 Sydney Street SW3 is now a four-bay, three-storey (plus basements) terrace in single ownership, facing east-north-east. The four buildings were originally one end of a long terrace, the rest lost during the Blitz.
When first inspected in 1978, the shop-fronts were distorted out of square, and the front ends of all the party walls between the shops were leaning progressively towards the right-hand flank wall. This was most out of plumb, at 150mm. Above first-floor level, walls were relatively plumb.
As bombing had removed the majority of the terrace, which caused the bookend effect, the prognosis for the surviving four bays was good. Re- inspection in 1994 showed that although the flank wall needed tying back for its own stability, the overall structure of the terrace had not apparently distorted any further.
A Victorian terrace
Nos 48-62 Bouverie Road, N16 is a medium length (eight-bay) four-storey terrace (including semi-basements) in multiple ownership, facing west- south-west.
When surveyed in 1988, the flank walls at both ends were leaning out. The lozenging of door and window openings matched the leaning flank walls. But there was little cracking visible in the front and rear elevations.
No 48, an end-of-terrace, was owned by our client. Its flank wall was 130mm out of plumb (and there were other unrelated internal structural defects).
As part of a comprehensive refurbishment, the flank wall was rebuilt plumb. The timber spine wall was replaced by masonry to buttress the party wall and provide a better bookend. Thus the end-of-terrace property was given a new lease of life.
A Victorian shop-fronted terrace
Nos 92-104 Fulham Road, SW3 is a medium-length (six-bay) four-storey (plus basements) terrace in multiple ownership, facing south-east.
When inspected in 1995 the end-of-terrace corner shop (No 92) had three circular cast-iron columns behind the fully fenestrated shopfronts on two adjacent elevations. The 3.7m-high columns tapered from 152 to 127mm at their heads, where they supported 230mm-deep x 280mm-wide timber bressummers that carried the structure above.
At some time in the past, the whole of the spine wall and approximately 60 per cent of the main rear wall of No 92 had been removed from the basement and ground-floor storeys to create an open-plan shop. Also, ties had been inserted between the flank wall and the internal superstructure, except where obstructed by the staircase within.
The columns in the corner shop were found to be leaning by up to 80mm towards the end of the terrace. The prognosis was for progressive future movement, which would have caused sudden column failure and collapse of the superstructure.
The columns were therefore replaced by steel 'picture frames' with moment connections for enhanced sway resistance. The frames were designed for piecemeal insertion using bolted joints. Flat jacks were placed under each end of the frames to prestress them to 70 per cent of the estimated vertical dead load. Sundry repairs and general tying together of the superstructure were also specified. While recognising that the cause of the damage could not be eliminated, these palliatives are expected to give a good lifespan.
Victorian conversion of Georgian shops
No 20-36 Baker Street, W1 is a 62m-long (nine-bay) terrace in single ownership, facing west-south-west.
This four/five-storey (plus basements) terrace was constructed around 1790 as houses, with open-front basement areas. With the rise of commerce during the 1800s, the ground floors were converted into shops by forming large openings in the front and spine walls and roofing over the basement areas. The structures above the shops are carried on bressummers which span between the party walls, sometimes with intermediate support from storey posts.
In 1996, the flank walls and all the party walls of 20-36 Baker Street were surveyed accurately for verticality. Small holes were drilled through the floors so that plumb-lines could be hung down the party walls. Small areas of plaster were removed from the party walls, so that plumb-line offsets could be measured directly to the brick faces. (In a previous case in East London we found modern plaster dubbed out 150mm to disguise distortion).
The results showed a consistent trend of the walls leaning towards either end of the terrace (see diagram). A few individual measurements were contrary to the trend. Also, the 'origin' of movement was not at the centre of the terrace but at approximately two-thirds of the terrace away from No 36. These variations from a 'textbook case' were attributed to lack of accuracy in the original construction.
The 14m-high flank wall of No 36 was leaning by approximately 150mm. Its margins of safety were significantly reduced, and without structural intervention the wall could have become unsafe within five to 10 years. It has since been inspected quarterly for further deterioration and its movement monitored continuously with electrolevels (electronic spirit levels), as used at Pisa by the bre.
It was hoped that the electrolevels would record the three components of structural movement in the bookend effect:
diurnal leaning in and out
seasonal leaning in and out
ratcheting, so increasing maximum displacement, year on year (see graph).
The electrolevels were validated by the Building Research Establishment1. The vast amount of data from hourly signals over 10 months was analysed by Alan Messenger of Flairline Properties, and results showed diurnal and seasonal cycling, the latter calculated at 14-28mm/annum.
Unfortunately the electrolevels were not sufficiently sensitive to detect ratcheting. The flank wall was 210 years old and approximately 150mm out of plumb. Averaged over the period, it would be moving out of plumb by about 0.7mm/year, too small an amount to be detected by the electrolevels.
Time's up for terraces?
These case-studies illustrate the wholesale distortion of the bookend effect, which can be seen in terraces throughout the uk. The life expectancy of terraces suffering from the bookend effect is poor.
Without structural intervention, progressive distortion will continue. Indeed it will accelerate as gravity acts more and more on the out-of- plumb flank and party walls and as old age reduces the robustness of the structural fabric. Eventually, the flank walls must collapse, followed later by each party wall in turn.
How much distortion flank walls can tolerate before their margins of safety and stability are eroded depends upon their height, thickness, bond, restraint, pre-compression and exposure. As a rule of thumb, walls between one-sixth and one-third of their thickness out of plumb require engineering analysis. Walls greater than one-third of their thickness out of plumb cannot usually be economically restrained by domestic-scale structural measures2.
To eliminate the bookend effect, vertical joints can be formed at intervals along the front and rear terrace walls to accommodate horizontal expansion and contraction. The joints should be located at party wall-to-external wall junctions to maintain lateral stability. Ensure also that the party wall is adequately bonded to the external wall. The frequency of joints can be less than for modern masonry3.
An alternative is to give the masonry a measure of tensile strength, so that in the cycles of expansion and contraction cracks are not allowed to develop. It is feasible to let horizontal reinforcement into the masonry by chasing out the bed-joints at the level of window aprons, and pointing in 2.5mm-diameter stainless-steel wire or proprietary twisted ss strip. However, the cost is likely to be greater than for installing vertical joints.
Unfortunately, as very few terraces are in single ownership, the owner of a mid-terraced property has little incentive to co-operate with these measures, or any other remedy.
Without eliminating or controlling the cause of the bookend effect, the following palliatives will increase the life of a terrace, though something is likely to yield eventually.
Build another property at the end of the terrace, to act as the bookend.
Tie the leaning flank wall to the internal structure of the end terrace property to increase its individual margin of safety - always recognising that the wall is tied to another part of the structure which is itself being displaced.
Rebuild the leaning flank wall, and buttress the leaning party walls by constructing the spine walls in more rigid construction.
For shop-fronted properties, brace or portalise the large openings.
The future of the terrace stock
Not all terraces suffer from the bookend effect. But examination of hundreds of terraces over the last 20 years, has more often than not found distortion by various amounts. South-facing elevations seem to suffer most from the bookend effect.
Some terraces apparently do not suffer at all. This anomaly may be due to particular combinations of materials and structural formats which provide compensatory effects. These include moisture movement4, the autogenous healing properties of lime-mortar and the visco-elastic behaviour of stressed brickwork5.
The prognosis for future life of Georgian and Victorian terraces is poor. Cures or palliative measures will have to be applied with increasing frequency in the twenty-first century. While old terraces have accepted 100 years or more of distortion, this has eroded their margins of safety and stability. Without structural intervention, end-of-terrace buildings will continue to bear the brunt of the bookend effect. They will progressively lean further and faster until they reach the point of collapse, followed in due course by each penultimate property in turn, like a row of falling books.
Clive Richardson is a technical director of civil and structural engineer R T James and Partners, and engineer to the Dean & Chapter of Westminster Abbey