Metrication fiascos #75: "L/100 km" is NOT a unit of anything

There is no definition of a quantity or unit for automotive fuel usage in the International Standard ISO 80000 Quantities and Units, so we should use national standards. The most common measures are the reciprocal quantities 'fuel economy' and 'fuel consumption':
1. The US Government mandates a new-car label stating 'fuel economy' with a unit of 'mpg'.
2. The European Union and Australia mandate a new-car label stating 'fuel consumption' with a unit of 'L/100 km'.

The latter is not a unit, by the international definition. It breaks the rule for forming compound units in ISO 80000 Part 1: General, 3.2 Combinations of symbols for units, which states that compound units are formed by multiplication and/or division of units (no mention of numerical factors). There is no other unit in scientific or public use which uses a numerical factor and breaks this rule. Furthermore, this 'unit' is invariably written 'L/100 km', which when parsed with standard mathematical precedence, yields a meaningless expression 'L/100*km' of dimension [length]⁴.

When petrol bowsers and car odometers were metricated and this 'unit' was introduced to Australia in 1976, I queried the Metric Conversion Board. The reply, in essence, said:
"1. The Europeans use this unit, so we should harmonise.
2. Most cars have a fuel consumption of 12 - 20 L/100 km, so their fuel economy (5 - 8.5 km/L) has unacceptable rounding errors when expressed with one decimal point."

I replied then, and I say now:
1. Adopting someone else's incorrect convention does not make it correct.
2. With improvements in technology, the average car fuel consumption will reduce, and one day it will be in the range 5 - 8 L/100 km, so the rounding error will then apply to this measure.

So what to do? Fuel consumption is the better quantity to use for comparing the fuel usage of two vehicles, since it can be subtracted to give their relative fuel usage. For the benefit of consumers, the world (including the USA, which is just about to revise its fuel economy label) should adopt it. We should write the unit as 'L/(100 km)' (as per NIST Special Publication 811 Appendix B5) so it can be parsed correctly by future 'unit-aware' software.