For convenience of interested parties, the Phototechniques web site lists:
Vol. 24, No. 5 Reciprocal Trade Agreement - Gainer
Vol. 24, No. 4 B&W Reciprocity Departure Revisited - Bond
These would be the July-Aug (No. 4) and Sept-Oct (No. 5) issues for 2003 if I'm calculating correctly.
To answer in part the questions Bob Carnie and Mike K have about adjusting development;
When you make a long exposure in the range where a film has reciprocity law failure, you'll probably have a range of a number of stops, probably a minimum of 5-7 stops or often much greater, especially in a night shot with manmade lighting and dark areas. If you took the darkest area and calculated the necessary exposure with reciprocity compensated for, then did the same with the lightest area, you'd find that the dynamic range (# of f-stops) of the scene has just expanded tremendously because of the differential in reciprocity failure between the light and dark areas. To compensate in part for this, development times are sometimes decreased to keep the contrast of the negative down, but that means marginally lower "film speed" as well. This gets out of hand rather quickly, and you often just have to accept some loss in highlights, shadows, or both. Or you can explore the stand development and other techniques used by the photographers who do cathedral interiors.
My advice is to bracket either side of your calculated exposures if the shot is important to you. I've found that bracketing with the Fibonacci series works well. That series is 1, 2, 3, 5, 8, 13, 21... each subsequent number is the sum of the previous two numbers, and I do this in either seconds or minutes. I do this because in my experience a half stop bracket is too little and a full stop is too much. It works for me in time exposures of under a minute up to a half hour or so. I started using the Fibonacci series when I was doing 2000 custom B&W prints a month using two hand-operated Omega D-V enlargers with no meter or analyzer. (Luckily I had a roller-processor for the paper.) I could eyeball the exposure and be right 95% of the time, and could always nail the second by going up or down to the next number in the sequence. It was also handy because I'd often get requests for reprints, and within the tolerance of the materials, I could always get a reasonably exact duplicate print without having written down exposures.
When you take the ratios of any two sequential numbers in the Fibonacci series, you get 1.618, which is coincidently, the golden mean ratio, and the number by which the log of the indicated exposure is multiplied in Mr. Gainer's reciprocity equation.