I started with the book in Microsoft Word format. I converted that to HTML, then used Dreamweaver to strip out the junk tags that Microsoft inserts. I spent some time inserting cross references, so that readers can jump from one section to another, and putting in images.

Then I ran the HTML file through Calibre (free ebook software) to produce the book in MOBI format, which can be uploaded to the Amazon DTP site.  You can use MobiPocket Creator instead – it produces smaller files, but I found that the layout produced by Calibre was closer to what I wanted.

The tables were the hardest part.  I tried to do them in simple HTML (using Dreamweaver) but unfortunately they don’t show properly on the smaller Kindles.   So I took screen shots of the training programmes, rotated them, converted them to GIF files, and put them in to the file as images.

Alan Jones has recently updated the age-grading tables, because the estimates of world-record times for older women were proving too soft. Apparently some females were obtaining scores over 100%. The men’s tables have not changed since 2004.

The age grading tables used on this website are a combination of Howard Grubb’s tables here for men, and women’s track, and the Alan Jones updates for female road records here.

You can download an Excel spreadsheet containing the age grading times used on the website here.

http://www.mcmillanrunning.com/mcmillanrunningcalculator.htm

http://www.hillrunner.com/jim2/

(Thanks to Andrea for sending these in.)

- the Serpentine Handicap (6975 metres)
- 20 miles
- 40 kilometres

There’s also an option (here) to calculate your age graded time based on any arbitrary distance, which also requires interpolation.

In the past, I’ve used linear interpolation. This basically draws a straight line between the observation before and observation afterwards:

Linear interpolation

IT = LT + (((ID – LD) / (HD – LD)) * (HT – LT))

IT = Interpolated time; LT = Lower Time; HT = Higher Time
ID = Interpolated Distance; LD = Lower Distance; HD = Higher Distance

But I’ve realised that isn’t very sensible when we have a better model, namely the Riegel Formula.

The Riegel Formula is HT = LT * (HD / LD)^n (where n is approximately 1.06)

I have adapted this to do non-linear interpolations of the age-grading formula. The formula calculates n by comparing the times above and below.

Rewriting the equation above:
n = (ln HT – ln LT)/(ln HD – ln LD)

Then we can use that n to calculate the intermediate time.

Non linear interpolation

Working from the lower bound:

IT = LT x (ID/LD)^((ln HT – ln LT)/(ln HD – ln LD))

or, working from the higher bound:

IT = HT / (HD/ID)^((ln HT – ln LT)/(ln HD – ln LD))

I’d like to thank David Knight for the use of his photographs.  And Ian Hodge for passing on the latest age-grading data.