Why the bookend effect is no myth
My article on the 'bookend effect' (aj 16.7.98) describes how certain end-of-terrace properties are pushed sideways, like bookends, due to the cyclical and progressive longitudinal expansion of the rest of the terrace.
John Mason's letter (aj 20/27.8.97) doubts that the bookend effect exists because his practice, Alan Baxter and Associates, 'has not found evidence of it nor does it know of any other engineer who acknowledges it'.
Galileo was subjected to the same fatally flawed logic. Nevertheless, I am pleased that Mr Mason has written, for he serves to highlight how easy it is for the bookend effect to be mistaken for more commonplace defects, mostly associated with the localised bulging or leaning of end- of-terrace flank walls, due to sulphate attack, eccentric bearing of floors and roofs, decay of embedded timber wall-plates and bonders, bomb blast, and wind suction.
If flank walls are being displaced due to any of these localised problems, then they usually remain vertical at their returns with the front and rear walls.
The subtle difference with the bookend effect is that the flank wall corners do not remain vertical, because it is the longitudinal thermal expansion of the front and rear walls which is pushing the flank wall sideways.
The bookend effect is not a localised problem of flank walls, but a problem of terraces as a whole. Frequently, there will be a combination of causes at play. Having diagnosed one or more commonplace local causes, it is very easy to overlook any wider issues, particularly if the client's brief relates to surveying one particular terraced property, and not the whole terrace. But of course the structure of a building extends to its very furthest walls, and not just the property boundaries in question.
There are three components of movement in the bookend effect, as illustrated by the graph in my article: (1) daily cyclical movement, (2) seasonal cyclical movement, (3) progressive displacement year on year.
Mr Mason does not take issue with the veracity of (1) and (2), but argues that the electrolevels mounted on Baker Street have not been able to show the third (progressive) component 'because no such movement is occurring'.
I stated quite clearly that the accuracy of the electrolevels was not sufficient to detect the progressive component in the time available; that is not the same as there being no component.
Electrolevels are a way of recording movement as it happens on line, via a pc. But other evidence of progressive movement is readily available to those who wish to observe it.
Long-term occupiers of terraces suffering from the bookend effect find that doors and windows have to be continually shaved as their structural openings distort, and gaps which open between floors and party-walls require covering with quadrant mouldings to stop the draughts (see sketches above).
These maintenance actions are promoted by the front and rear elevations being racked sideways, causing lozenging of door and window openings. Such distortions do not occur 'quite soon after they were built', as Mr Mason maintains, but slowly and progressively due to creep distortion of the masonry.
Logic dictates that terraces were not built with lozenged openings, nor party walls leaning so consistently to either end, nor flank walls out of plumb by 150mm or more. The obvious answer is that they have distorted progressively over 100-200 years.
The last remaining strand of Mr Mason's reasoning is that 'there should be progressive leaning-out of each party wall towards the end of the terrace' which, alas, is not shown by the measurements in the Baker Street case study.
If the measurements are studied closely, two other deviations from a textbook case can be seen: the neutral point of party-wall movement is not at the centre of the terrace, but about two-thirds from one end, and also one storey of one party wall is leaning the 'wrong' way.
These three deviations show the vicissitudes and inaccuracy of Georgian party-wall construction, but they do not detract from the fact that there is a remarkably consistent trend of displacement towards either end of the terrace.
The other three case studies also record measurements of flank walls and party walls consistently leaning the same way. I chose four case-studies to illustrate the range of vulnerable buildings, Georgian and Victorian housing and shop-fronted properties, but Mr Mason can rest assured that I have studied hundreds of examples over the last 25 years. A review of his 1000 allegedly unaffected properties might prove quite illuminating!
Lastly, I turn to Mr Mason's stated reasons for writing to you. The first reason is erroneous, as he misrepresents my article; I clearly stated that not all terraces suffer from the bookend effect.
The second reason is his belief that one of my six suggested repairs is likely to do more harm than good, ie cutting movement joints at party- wall lines. The details of the joint are similar to those that are accepted and widely used for modern cavity-wall construction. The joint is formed on one face of the party wall, so that a fully bonded return between the party wall and one half of the external wall is maintained. This stable return is used to restrain the other half of the external wall via sliding dowels.
We have used the detail successfully for a number of years and are happy to supply your readers with working details upon request. If Mr Mason does not like the repair he can use one of the other five. It is certainly not a good enough reason to remain in a flat world.
R T James & Partners