Despite its title, this book is mainly about Roman wooden force pumps, rather than Roman pumps in general. It presents a wealth of detail and engineering analysis of the 18 known Roman wooden pumps, based not only on published works but also on unpublished reports and archive material and, as far as possible, on first-hand observation of those finds that have survived. The main argument and discussion is set out in six short chapters (or ‘Sections’), occupying in total 67 pages; these are followed by a series of appendices, one of which in turn is a detailed set of ‘Reports’ (with their own appendices) on each of the wooden pumps.

The double-acting force pump was invented by Ctesibios in Alexandria during the third century BC, and was originally constructed of cast bronze parts. By the first century AD the design had been radically re-engineered to enable its construction in cheaper materials, with the necessary piston cylinders and connecting chambers hollowed out of a single block of oak, and fitted with one-way inlet valves. The pistons were made of wood, shod with iron and driven via connecting rods (‘con-rods’) linked to a rocker arm; the con-rods enabled a pump at the bottom of a well to be worked by an operator at the surface. This re-engineering of an originally bronze design in wood will have lessened the cost of both materials and labour, and should have made the device more widely affordable.

The 18 known wooden pumps all come from Italy, Gaul, Germany and Britain, and range in date from the first to the fourth centuries AD. The majority (13 of 18), found in wells with their liners intact, were evidently used in those wells, whose depths range from 2.8m (Benfeld) to 26.4m (Tarrant Hinton). A fourteenth pump was also found in a well, another in an aqueduct shaft (but which probably was used as a well, see below), another in the basement of the Trier amphitheatre and one, from a modern construction dump, lacks its original context. Stein correctly, I now believe, rejects the interpretation of a find at Nocelli as a pump.

One of the most important contributions of the book is the calculations of lift, discharge and energy requirements using a common approach with a consistent set of assumptions. Output seems to have ranged from roughly 15–40l per minute at a lift of 10m—in one case, even up to 26m (Tarrant Hilton, output calculated at nearly 25l/min)—if the pumps were worked at 15 cycles per minute.

Stein repeatedly asserts his view that many, even most, wooden pumps were used “at ground level” (pp. 15, 20, 25, 65 & 67), by which he seems to mean that they were used for firefighting. Little evidence is presented in support of this view, which appears to derive from the fact that the ancient texts referring to pumps (usefully tabulated, pp. 26–28) mention them principally in the context of either fighting fires (10 texts), or spraying perfume on the crowd in the theatre or the amphitheatre (at least 6 texts), with only one explicitly referring to pumping water from wells. But the texts do not specify whether they are talking about wooden or bronze pumps; and the uses that they describe largely reflect the contexts in which such pumps impinged on the consciousness of a literary elite, rather than the true extent of their use. Stein himself notes that wooden pumps are “heavier and less portable, and their materials are inherently less durable, than those of bronze pumps” (p. 21).

Undue weight is given to the idea that the treatment of the Bertrange pump was ‘aesthetic’ and that this indicates it was originally used at ground level before being installed in the well in which it was found; the scenario envisaged for its purported original use is, however, entirely speculative. The discussion of the Colle Mentuccia pump, found in a shaft apparently related to the Aqua Appia in the area of its springs, goes more seriously astray. Stein takes the reconstruction drawing (reproduced as figs 68 & 74) showing a rocker arm literally, not seeing that it is a purely conjectural reconstruction (as fig. 69 makes clear), and thus he believes that the rocker arm was mounted directly above the pump and worked from the bottom of the shaft. As a result, he is puzzled as to why, on the drawing, the rocker arm allows the pistons very little travel, and also how and for what purpose the pump was used if it was operated from the bottom of the shaft. But the drawing is a conjectural and erroneous reconstruction; there is no actual evidence for the configuration of the rocker arms, nor that the pump was worked from within the shaft, and it was surely operated from the surface via con-rods in the normal way, when the aqueduct shaft was opportunistically (re-)used as a well. Again, this is a fixed pump and not a portable one (cf p. 51).

There is in fact no evidence that any of the wooden pumps discussed were ever used at ground level; even the Trier amphitheatre pump, which may have been a portable pump for spraying perfume, could also have been a fixed installation for the same purpose (as mentioned by Seneca, Quaest. Nat. II.9.2). The obvious alternative to Stein's view is that the bronze version of the pump (still used in the second century AD, and probably also in the third century) continued to be used as a portable pump, for fire-fighting and for spraying perfume in theatres and amphitheatres, while the heavier, less portable but more affordable, wooden pump was primarily used in wells.

The book is not easy to use, and is poorly designed, in both structure and typographic layout (overlong line lengths, cramped headings). The notes are sandwiched in between the main text and the appendices; the bibliography is hidden away as Appendix K in the middle of the book; illustrations are relegated to the back. Important material is buried in the appendices; for example, the reader interested only in the main text would still be well advised to read Appendix M, a summary of the main features of each pump, before proceeding very far with Section 3 (‘The known wooden pumps’), as it provides key information that properly belonged in that ‘Section’.

Despite these reservations, there is much that is useful in this book. In bringing together the data and illustrations for all of the known wooden pumps and presenting new calculations, it is a considerable advance on previous discussions, and it will become the standard reference work on the subject.