What is the waist-to-arm ratio for?

It is the cheapest partition signal available. Waist circumference tracks abdominal fat closely and is barely affected by training, while arm circumference tracks lean mass. A bulk where the ratio holds or falls is partitioning well; one where it climbs is adding more fat than muscle, and that shows up here weeks before it shows up anywhere else.

How accurate is the Navy body fat estimate?

It is a circumference-based approximation with meaningful error against a DEXA scan, and Base does not present it as anything else. Its value is as a consistent series rather than an absolute — measured the same way each time, its direction is informative even where its absolute number is off by several points.

Fat-free mass index — lean mass normalised for height, the same way BMI normalises total mass. It is a more useful number than BMI for anyone who trains, because it does not penalise muscle. It needs a body-fat estimate as an input, so its error inherits that estimate's error.

Why do some numbers show a percentile and others do not?

Because only some have a published distribution behind them. Where a reference publishes percentiles, Base shows your rank. Where the reference is a clinical or consensus range with no distribution, it shows the band and explicitly no percentile — inventing a rank there would be fabricating data.

How often should I measure?

Every two to four weeks. Circumference measurements have real measurement error from tape placement and tension, so measuring weekly produces a series where the noise exceeds the signal. Monthly is enough to catch a partition problem while it is still cheap to fix.

Track Body Measurements and Composition

Waist-to-arm

The cheapest honest partition signal there is.

waist 32.0 in ÷ arm 15.5 in = 2.06

four weeks later
waist 32.9 in ÷ arm 15.6 in = 2.11   → rising: fat outpacing lean

Waist circumference tracks abdominal fat closely. Arm circumference tracks lean mass, imperfectly but usefully. A bulk where the ratio holds or falls is partitioning well; one where it climbs is putting on more fat than muscle.

This matters because it feeds the rate controller. The training-age tier Base starts you on is a prior, and evidence moves it — a lifter adding strength quickly with a flat waist is partitioning well and earns a higher allowance than the table gave them. Waist trend is one of the inputs to that judgement, alongside trend weight and strength.

The three composition figures

Figure	What it is	Honest limitation
Body fat	Navy circumference estimate	Real error against DEXA; useful as a series
FFMI	Lean mass normalised for height	Inherits the body-fat estimate's error
BMI	Total mass normalised for height	Does not distinguish muscle from fat

BMI is included because it is a clinical reference people encounter, not because it is a good measure for a trained person. It cannot tell muscle from fat and will call a lean, heavily muscled lifter overweight.

FFMI is the better number for anyone who trains, because it normalises lean mass rather than total mass. Its weakness is that it needs a body-fat estimate as an input, so its error inherits that estimate's error.

Navy body fat is a circumference approximation. Its absolute accuracy against a DEXA scan is modest, and Base says so rather than presenting it as a measurement. Its value is directional: measured the same way each time, the trend is informative even where the absolute figure is several points off.

That last point generalises to this whole screen. Consistency of method beats accuracy of method when what you want is a change over time.

Percentiles, and where they are absent

Long-pressing any of these opens its explanation, which includes how you compare — and this is where the app is unusually careful.

Base renders exactly one of four things and never a fifth:

A percentile strip, when the reference publishes a real distribution
A band with no percentile, when the reference is a clinical or consensus range
An explicit statement that there is no population comparison for this number
What is missing from your profile that would enable one

The second case is the important one. A clinical range like "normal BMI is 18.5–25" has no distribution behind it, so producing a percentile from it would be inventing data. Base shows the band and says there is no rank rather than fabricating one.

The fourth explains why a comparison sometimes does not appear: without sex or birth year, there is no cohort to compare against, and Base refuses to assume rather than defaulting to men's norms.

Composition figures with a per-site log underneath. Each figure long-presses into its reference band, which is the only context that makes any of the three mean anything.

Measuring consistently

Circumference measurement carries real error from tape placement and tension, and that error is what decides how often measuring is worth doing.

Same landmarks each time — navel for waist, mid-bicep flexed or unflexed but always the same
Same tension, snug without compressing
Same time of day, before food
Same person measuring, if someone helps

Every two to four weeks. Weekly produces a series where the noise exceeds the signal and invites reacting to tape placement.

Photos are the other half

Measurements catch a partition problem numerically; progress photos catch what a tape cannot. Together they are a far better instrument than either alone, and both are far better than the scale on its own — trend weight tells you how much changed and nothing about what.

Common mistakes

Measuring weekly. Noise exceeds signal.

Measuring at different landmarks. Two inches up the arm is a different number entirely.

Treating the Navy estimate as a measurement. It is an approximation, and the app says so.

Reading BMI as meaningful for a trained person. It cannot see muscle.

Measuring the waist flexed or after a large meal. Both add inches that are not fat.

Ignoring a rising waist because the scale rate looks fine. The scale says how much; the waist says what.

What to do next

Measure your waist and one arm today and note the ratio. Repeat in a month — if the ratio has climbed while your trend weight has risen as planned, the surplus is too large regardless of what the rate says.