Recently, Allen Edmonds had its home page redesigned, and now includes a shoe fit guide with sizing recommendations. Naturally, this was a perfect opportunity for me to compare them to Sizeadvisors' calculated sizes (available under 'Your Profile'): Would it differ greatly? Whose calculation is more accurate? Here's the story…

### The Allen Edmonds shoe sizing guide

Allen Edmonds' shoe sizing guide is a PDF that comes in three pieces: The first page provides the upper part for measurement, and measuring instructions; the second page provides the lower part for measurements, and instruction how to put the first parts together; and the third page adds a conversion chart for US, UK, and EU sizes, and a width chart.

For such a chart to work, printing it at the correct resolution is paramount. Allen Edmonds' chart provides a convenient method to check by having a special field for a credit card: If your card fits, the chart was printed in the correct resolution.

Another way to test it, though, is to check the difference between two sizes. Since the US system is build upon the UK system, one full size difference equals 1/3 inches. Consequently, the differences in length between a (US) 4 and a (US) 16 are 12 units – that's exactly 4 inches. I'll come back to this is a moment.

Usually, one is supposed to glue the first to parts together but I dispensed with that, since I only wanted to figure out how the chart is constructed. Cutting and gluing also might have introduced errors.

Therefore, I took three different measurements of the relevant pieces. First, the length of the print on the second page: From the back of the heel to the end of the print, it measures 7 inches. Let's call this the *base length*.

Second, the differences between each sizes printed on the first page: All units should have a difference of 1/3 inches, and that was indeed the case. Finally, the length of the lowest size, (US) 4: It's a tad longer than 13/8 inches. My measurement tape was not that precise (in Millimeter, it was about 415), I decided to add 1/32 inches, yielding a total length 53/32 inches for a (US) 4.0 size. Consequently, a (US) 5.0 is 1/3 inches longer, a (US) 6.0 is 2/3 inches longer, etc. Let's call this the *variable length*.

With these data points, it was easy to reconstruct what shoe sizes Allen Edmonds recommends for each foot lengths:

US size | Base length (in) | Variable length (in) | Foot length (in) |

4.0 | 7.0 | 1.656 | 8.656 |

6.0 | 7.0 | 2.323 | 9.323 |

8.0 | 7.0 | 2.990 | 9.990 |

10.0 | 7.0 | 3.656 | 10.656 |

12.0 | 7.0 | 4.323 | 11.323 |

14.0 | 7.0 | 4.990 | 11.990 |

16.0 | 7.0 | 5.656 | 12.656 |

In other words, you should wear shoes in (US) 10.0 if your feet measures up to 10.656 inches but not less than 10.489 inches – for that would be a (US) 9.5.

Looking up your shoe width is a bit more problematic. There are 13 so-called “zones” labeled W1 to W9, and WA to WD, with a total difference from the lowest to highest zone of 63mm, resulting in a difference for each step of 4.85mm. The most narrow zone ('W1') starts at 69mm, the widest zone ('WD') ends at about 132mm.

Because shoe widths depend on shoe sizes, you're then supposed to look up the shoe width for a given zone in the width chart. For instance, if your feet measure anything between a (US) 7.5 to 9.5 and they're in the zone W7, then you're a 'D' width. For the same sizes, you're an 'E' if your feet width are in the zones W8 or W9.

This may sound confusing, at first, but makes sense if you have the guide handy.

Now, it's one thing to re-construct Allen Edmonds' recommendations; what's interesting is what underlies their decision to recommend these sizes. Why these and not others?

### Insights about widths

Before I look at the shoes sizes, let me first point out a surprising result about Allen Edmonds' width recommendations.

Usually what's called “width” in the U.S. (and “fitting” in the UK) is said to be the ball girth of the last. Each unit – for instance, a step from D to E for a given size – is said to be an increase of 3/16 inches in circumference. While the true width of a last should therefore increase with each unit, it should do so in a smaller degree.

If the theory is correct, one should thus expect less than 3/16 (or 0.1875) inches for each unit in Allen Edmonds' sizing guide. The data, however, contradicts the assumption. Actually, each step is larger, namely 4.85 mm (or 0.191 inches)!

The contradiction becomes even worse when translating these width zones into actual shoe widths. For one shoe width may even cover two zones. For instance, you're a (US) 9.0-B if your feet are in the zones W4 and W5. You're a (US) 8.0-D if your feet are in the zones W6 and W7. You're a (US) 11.0-E if your feet are in the zones WA and WB.

In other words: For certain length, people with feet that differ up to 9.7 mm in width are meant to wear the same size! That's **nearly a centimeter**. You may want to keep this in mind when looking at the measurements of other members in the detail view for an ideal fitting.

That said, let's now look at the shoe sizes.

### Reverse-engineering Allen Edmond's recommended allowance

As mentioned in earlier posts, I believe shoes sizes are made for one thing, only: To make sure you have a reasonable amount of free space in front of your toes when wearing shoes. I call this nominal allowance, for it ought to be constant within a particular sizing system.

It's called nominal since it's derived from the “name” of the length, ie. the shoe size. There may be extra free space due to the styling decisions of the last maker – the formal allowance – but this varies for each last, and is therefore negligible.

What nominal allowance does Allen Edmonds recommend? Given the sizing chart, it's easy to work backwards to figure it out: Any shoe should have a nominal length (a minimum length in inches or centimeter for a particular shoe size) to accommodate your foot length, and the nominal allowance. We only need to rearrange the shoe sizing formula for US mens sizes:

Nom. length (in inches) = 1/3 * (us_size + 24)

For instance, all shoes in (US) 9.0 should always have an inner length of 11 inches minimum. We only need to do this for each size, and subtract the calculated the foot length from above. Here's the table for a sample of sizes:

Sizes (US) | Nom. length (in) | Foot length (in) | Nom. allowence (in) |

4.0 | 9.33 | 8.656 | 0.677 |

6.0 | 10.00 | 9.320 | 0.677 |

8.0 | 10.67 | 9.990 | 0.677 |

10.0 | 11.33 | 10.656 | 0.677 |

12.0 | 12.00 | 11.320 | 0.677 |

14.0 | 12.67 | 11.990 | 0.677 |

16.0 | 13.33 | 12.656 | 0.677 |

In other words, Allen Edmonds recommends a nominal allowance of slightly more than **2 full units** (0.677 instead of 0.667 inches). The difference is quite small, so it's probably due to a rounding error, mis-measurement, or a glitch made by my printer.

This is close to the expected value: Sizeadvisors uses a nominal allowance of 1.5 units, or 0.5 inches, to calculate your US shoe size. This is slightly less than Allen Edmonds' recommendation.

Which one is right?

There's no way to say, strictly speaking. While most pediatricians and physicians seem to agree that some shoes are too short for one's feet, and others are definitely too long, there's some disagreement what the best is, or should be. However, don't forget that we're speaking about a difference of a mere 1/6 inches! In general, people usually have other problems when trying to find shoes that fit.

What we can do, though, is to check whether the fitting experience of Sizeadvisors' members is consistent with Allen Edmonds' recommendations. Let's see…

### Putting the recommendations to a test

I used the same method as before, since Sizeadvisors' measurement guide is similar to the one of Allen Edmonds. First, I calculated the recommended shoes size for each member's maximum foot length. Second, I've subtracted this from the size of shoe by Allen Edmonds rated “3” or “4” by members.

Consequently, a zero difference indicates that a member rated a pair of Allen Edmonds shoes in Allen Edmonds' recommended size to fit fairly good, at least. This is the distribution of all results:

As usual with bars like that, the area of a bar indicates the relative frequency, not its height. Each bar has a width of 0.5 units, thus you need to divide the numbers on the left by two.

As you can see, over a quarter of all shoes being rated do not differ from Allen Edmonds' recommended shoe size. This, in turn, also means that close to three-quarters of all shoes being rated to fit fairly or really good, do differ to the recommended size. In fact, a large majority – close to a third – of all shoes being rated are actually half a size smaller.

Since the difference can also be expressed in shoe units for the allowance, one could also say that our members prefer an allowance of 1.5 instead of the 2 units for most of their shoes. Of course, this doesn't invalidate Allen Edmonds' sizing guide. The data at Sizeadvisors is not a proper random sample, so there could be a systematic bias in the data.

One possible reason, for instance, is the prominence of Allen Edmonds' 5-65 last at Sizeadvisors. According to Allen Edmonds, it's their longest last. If you happen to follow the discussion in the English shoe forums, you know that many consider it also their most narrow last. Thus, a common recommendation is to go down half a size and up one width for this last.

Could it be that this caused the previous results? If that's the case, we should see the ”-0.5” bar losing support if we remove all ratings of shoes made on the 5-65 lasts. Here's what happens:

This chart shows rated sizes differ to Allen Edmonds recommended size without the 5-65 lasts. Contrary to expectations, we see the “0” bar lose support. The ”-0.5” bar (Sizeadvisors' calculated size) even gains support! In other words, Sizeadvisors' calculated size becomes even more accurate that Allen Edmonds' recommendations.

Why is that? Since the only difference are the ratings made for shoes on the 5-65 last, that needs to be the cause. Here's the same chart for all rated shoes on 5-65 lasts:

Finally, here Allen Edmonds sizing guide is more accurate than Sizeadvisors' calculated size: The majority seem to prefer a nominal allowance of 2 units for the 5-65 last, on average. This is contrary to common shoe forums recommendations, and a rather unexpected result. I'll look at the details in another post.

### Conclusion

Allen Edmonds' shoe sizing guide is a seriously needed help for any shoe buyer. Although it differs to Sizeadvisors' calculation, the difference is within reasonable limits, if you consider that there seems to be no proper scientific studies on the ideal allowance.

However, one problem seems worth mentioning.

One should really ignore the conversion chart on the third page. There, we're told that the US to UK conversion is only half a size for US sizes 6 to 13, an, interestingly, and a full size for US 14 to 16. This makes no sense, unfortunately.

The construction of the shoe size scales indicates a difference of one full size between the US and UK systems. All considerations support the conclusion: the maths, the internal consistency of recommended allowances, earlier studies, and the ratings here at Sizeadvisors.

That aside, it's great to see such an influential company try to get its customers buy shoes in the right size. As the above charts also show: Some people really need to update their idea of a good fit.

Allen Edmonds' sizing chart is a first good start.