How to specify: A century of works on building specification works

1840-1900

Possibly because of the rise of the general contractor (a new phenomenon, of which Thomas Cubitt's London-based company was the first, in 1815), books tackling specification writing and other aspects of contract documentation were published in Britain and the USA from the early 19th century (see list at the end of the article). An early example was Alfred Bartholomew's Specifications for practical architecture, a popular work which saw at least four editions (1840, 1846, 1886 and 1893), the last two under Frederick Rogers. Clearly there was a need for such a book and, at least to begin with, little competition.

The book was in two parts, a general introduction, and then a set of his own model specifications with numbered clauses. His advice was good, though the language a little obscure to modern readers:

In other words, be precise, concise, unambiguous and clear, or the builders will get you! But this is not easy (evidently). To make it easier, Bartholomew had an interesting suggestion:

He was advocating a clause database. One example of multiple qualifiers was "… glass, properly bedded, bradded, and back-puttied". He didn't mention subtracting the bits that aren't 'requisite' – perhaps he should have – it is just as important. His book contained the fruits of this approach in Part II.

"I beg to give one strong recommendation, which is, never attempt to describe in one paragraph, several things of different qualities; for the exceptions and the qualifyings which such sentences require, render them both more troublesome to compose and refer to; while for the saving of only a few common words, great ambiguity, if not contradiction, is the almost constant result."

He was arguing for breaking up lengthy rambling paragraphs covering several topics into a series of separate paragraphs, one per topic, even though some repetition may occur. This is still good advice.

"The turn of a phrase, the situation of a single word, the causing or the avoidance of a possible ambiguity, may sometimes involve the question of many hundreds and of even many thousands of pounds; while it ought to be the duty of the professional man, to take care that when a contract is entered into, no disappointment may ever occur, as to how much work the builder has to perform, and how much money the employer shall have to pay for that work."

So, the extra effort he advocated was worth it, at least to the 'professional' man. Part I contains much more advice along these lines, still sound today.

In Part II, Bartholomew offered a collection of his own clauses, numbered for easy cross reference, for a series of projects e.g. "A specification for erecting and completely finishing fit for use and occupation, a Dwelling House of the 2nd (or 3rd) Rate …" ('Rate' being defined in the building regulations). Guidance was included, in brackets. Even if readers did not use the text offered, they were encouraged to use the same method. He also introduced blanks, so readers were forced to insert their own values – reusing those he had used would not do.

1000. To construct for the support of the sleepers of the ground-flooring, brick piers not more than three feet apart, each 9 inches square, 9 inches high, and with the addition of a foundation 6 inches high and 13½ inches square.

1001. To construct a barrel drain 12 inches bore from [insert] to [insert] consisting of 4-inch brick-work, and stuccoed on the inside over the lower half thereof ¾ inch thick with pure quick Parker's cement; …

Later examples simply referred to the numbers of previous clauses – that is, his clauses comprised a clause library, from which the specifier could select, just as Bartholomew himself did. For example, between new clauses 1093 and 1094, we find "Rubbish, &c. (See § 989)", and at the end of 1094 we find "(See §§ 990-1)". This reduced bulk and made clear that this referenced material was standard text. There was no need to continually reinvent the wheel.

Blanks, clause numbering, a database of clauses to be reused or modified as appropriate, even key words (in the margins) – all are parts of the modern master specification. Bartholomew was ahead of his time.

Thomas Donaldson's Handbook of specifications (1860) was published as a successor to Bartholomew's work:

"The volume of Specifications by the late Alfred Bartholomew being out of print, and a work of the same nature being generally demanded, it was at first conceived, that a reprint of that useful work, with modifications and additions, would supply the desideratum. … [but] … the errors of taste and construction therein alluded to are confined to the inferior practitioner; and such a revolution has been effected in various operations, that the professional man required a work more in accordance with the improvements of the age …"

Clearly, the need for such a book continued to be felt, and indeed it has not abated since. That specification practice had improved must have been thanks, in part at least, to Bartholomew's efforts. Just as Bartholomew's book had two parts – introduction and suggested text – so did Donaldson's. From his introduction:

"Where Builders of high established character undertake a work, great minuteness of description may not be necessary; but the Architect cannot be too elaborate or cautious, when having to do with a stranger or person of doubtful reputation, as sometimes happens in the case of open competition for public bodies. These boards are less cautious as to the character of the Contractor than anxious to secure the lowest amount; trusting to remedy any defect by the stringency of the contract and elaboration of the specification; forgetful that, even with these precautions, a wide door is still left open for fraud and bad work by unscrupulous men."

Things haven't changed much – documentation for public works still needs to be more thorough than for private sector projects, thanks to much more open tendering, much less connection between documentors and constructors, and lowest-price tendering, which is still prevalent in spite of all the difficulties it brings, including fraud and bad work.

Donaldson suggested a 'skeleton specification', under the following trade headings:

Life was simpler then – only 15 work sections. We recognise most of these today, as work section titles, though 'Smith and bellhanger' has disappeared (the bells were used for summoning servants). Donaldson gave general technical advice under each of these headings, for both new work and repair, before moving to the second part of his work.

Here, Donaldson's approach differed from Bartholomew's – he didn't offer his own text, rather he uncritically provided 46 project specifications, which readers were left to mine for themselves. Critical comment was probably seen as impolite – looking a gift horse in the mouth, so to speak. Readers might have simplified the task of mining this material by referring to specifications authored by names they knew and trusted. Hopefully they would have chosen on the basis for their reputation for fault-free construction, rather than award-winning design. They might also have selected their sources by considering like projects, or on a regional basis.

Specifications provided include works by luminaries such as Charles Barry (Houses of Parliament), George Gilbert Scott, Thomas Henry Wyatt, James Bunstone Bunning, R. Stephenson (High Level Bridge, Newcastle) and John Dobson (Railway Station, Newcastle). While they may have been difficult for specifiers to put to practical use, these specifications are a mine of information for the historian of 19th century British construction.

Interestingly, Bartholomew's work was republished later in the century as has been noted, even though Donaldson had sought to replace it. Presumably readers preferred the numbered text approach, the library of clauses, the inserts, the consistent approach. That is, they preferred the master specification.

1900-1940

Publication of works in this genre continued into the 20th century (see list at the end of this article). But writing about specifications had also moved beyond mere edification of 'professional men', and into formal education.

The International Correspondence Schools Question Paper: Specifications (1900) asked 48 questions. Though students are the audience here, this does not mean that they were school leavers or undergraduates. Rather, the corresponding students were probably building professionals of various kinds seeking a formal qualification in specifying, as an aid to career development, or simply seeking to bridge a gap in their previous education.

So what sort of questions were asked? Here is a sample:

(1) What is a specification?

(9) Name two words the use of which should be avoided, unless their meaning is fully explained.

(14) What is meant by debris and rubbish?

(18) (a) In what connection is the word provide used in a specification?

(b) When should the phrase provided and set be employed instead?

(22) Why does the specification require the finished floors to be covered with felt paper?

(28) What test is applied to the plumbing system, prior to the final acceptance of the work?

(32) How is the veranda lamp controlled?

(36) In what contract would the tin speaking tubes and attachments be included?

(39) At what points on the building are galvanized-iron wire baskets placed?

(47) Specify the painting of a veranda with three coats of paint, including floors and natural-finished ceiling.

The answers were not included, unfortunately, so we must rely on our imagination to provide the two words which should be avoided (I can think of several).

Richard Kirby's The elements of specification writing – A text-book for students in civil engineering (1913) shows that it was not just architects, or building specifications, that were getting the attention. The advice was good – one wonders why so many specifiers still don't get it right, given that good advice has been around for such a long time. Kirby covered tendering and contracts, as well as specifications, and so only three chapters are of interest here: one on clearness and fairness, one on methods of specifying, including brands and methods of construction (both methods deprecated), and one on specific clauses (e.g. for a small earth dam).

On clearness, he recommended several points, all still sound:

• Prepare an outline (helps prevent repetition etc.)
• Do not use long sentences
• Use commas sparingly
• Avoid the use of many-syllabled or high-sounding words
• The use of pronouns, especially of relative pronouns, should be reduced to a minimum
• Give directions, not suggestions
• Aim to make your language crisp and concise.

He wouldn't have been an engineer if he didn't have a go at architects:

"Architect's specifications are often so abbreviated as to contain very few complete sentences. "Plaster on hallways to be …," etc. Fortunately the practice has not extended to engineering specifications."

Tom Thumtack's article, 'Specifications' (1914) was amusing more than really instructive, with one civil specification recalled as simply "This dam is to be built like the first dam only a damn sight better", but augmented with the mailed fist of the contract administrator. The lesson being, I suppose, that you can be too succinct, though Thumtack was here rueing the 'good ol' days'.

He did, however, following Bartholomew, flag the distinction between standard text and project-specific text, which finds its formal expression in two-part specifications like NBS Scheduler and NBS Domestic Specification. Less formally, this is the basis for all master specification systems, which in effect develop and maintain the 'cards', and help the specifier with the 'special instructions' (e.g. by providing inserts and guidance):

"We use the card system in the office. So much of every specification is common to all and so many materials are specified year after year in exactly the same manner that our specification-writer simply makes notes from his index and the stenographers find the cards to which they refer, and copy them. Only the special instructions are dictated."

Horace Peaslee's article 'Streamlined specifications' (1939), which drew on Thumtack's article, has been very influential, as anticipated by its publishers:

"It is very gratifying … to be able to introduce … the first radically new development in the writing of architectural specifications within the memory of the oldest practitioner."

Master specifications the world over use the streamlining system first suggested by Peaslee, making full use of key words and colons:

"With the body of the specification left free for technical details only, the specification writer may then express his requirements in clear, concise form in headings and subheadings, without sentence structure, using phrases in preference to clauses, with only essential adjectives or adverbs, with no articles, definite or indefinite, unless positively required."

One simple before-and-after example offered by Peaslee was as follows. Readers will recognise both – the full (à la Kirby) 'before' version still appears in many DIY and engineering specs:

"Waterproofing shall not be applied at temperatures lower than 50 degrees F."

This full sentence is reduced to:

"Temperature for application: 50°F. - minimum."

We are now in modern times, when succinct really means succinct. Kirby must have been turning in his grave (if he was in it then). Streamlining took off quickly in the USA, with 11 articles on the subject in Progressive Architecture between 1946 and 1949, for example, and legal imprimaturs from a committee of housing attorneys, the counsel for the AIA's New York chapter, and the counsel for the AIA (Small, 1949). It goes without saying that streamlining, as an editorial technique, can only improve an already technically decent specification; it cannot convert a technically faulty specification into a decent one.

The essentials of good specification practice

The essentials of good specification practice haven't really changed since at least 1840. For example, those preparing advice for specifiers have consistently advocated conciseness. However, 'concise' has been a relative term. What was regarded as concise in the early 19th century wouldn't have passed muster under Kirby, and his ideas don't meet today's standards.

Interestingly, master specifications, with standard structure and clauses and inserts for project-specific text, were anticipated very early on, long before any commercial subscription systems were actually published. Probably works such as Bartholomew's, overhauled just once a decade or so, were sufficient 'masters' before the mid-20th century.

But things are moving much faster now – standards publishers are active (ASTM was founded in 1898), new products and construction technologies are an everyday occurrence, society's expectations are continually shifting, and building codes are perpetually being revised. Maintained (and inevitably commercial) master specification systems with 6 month updates, or better, are an essential consequence – the modern version of the second part of Bartholomew's work.

And, though the principles remain much the same, a continuous stream of books on general specification practice is required to take care of the first part. Let’s just hope some specifiers actually read them. The writing of specifications is still not taught to any great extent at undergraduate, or any, level – in spite of the best efforts of pioneering organisations like the International Correspondence Schools. Specifiers, by and large, must train themselves.