Why can I not record my cat’s weight as five pounds?

Recently, when surfing the internet, I came across some correspondence between a cat lover and a supplier of cat food. When the cat lover wished to order cat food the supplier’s web site required that he enter the cat’s weight in kilograms. However, he wished to enter the cat’s weight in pounds. He went on to say that he had measured his cat’s weight by first weighing himself (13 st 4 lbs, or 186 lbs) and then weighing himself while holding his cat. The rest of the article was devoted to correspondence between himself and the cat food company in which he voiced his preference for using pounds.

This correspondence highlighted a sequence of bad practice so I will not embarrass the original writer by naming him or identifying the web site.

First, the use of metric units is standard practice in both the veterinary and medical worlds. Information aimed at vets and medical practitioners rather than the public pertaining to drugs is sometimes available only in English even though the products might be sold in many different countries. In such cases, all information relating to the animal’s or person’s dose is only given in metric units. If the public insists on using imperial units to express their own or their pet’s weight, they are adding an additional source of possible error: in the IT industry there is a saying “bugs breed in the corners and congregate on the boundaries.” Converting between different units of measure is one such boundary and, in this case, an unnecessary boundary.

The original writer’s second source of error was to weigh himself with and then without his cat and to assume that the cat’s weight is the difference between the two recorded weights. The writer did not specify the type of weighing device that he used, so I will assume that it was the same as my own bathroom scales. My scales have a four-digit display and can switch between kilograms and stones and pounds. Weights up to 100 kg are displayed with a rounding factor of 0.2 kg and above 100 kg the rounding factor is 0.5 kg. In imperial mode, the rounding factor is 0.5 lbs for weights up to 10 st, but above 10 st, the rounding factor is 1 lb. The observant reader will note that 0.2 kg and 0.5 lbs are about the same as 1 lb and 0.5 kg. Thus, between 10 st (63 kg) and 100 kg, my scales have twice the accuracy when displaying metric units compared to imperial units. Our cat lover weighs 13 st 4 lbs, so his weight falls into this area.

If his weight was accurate to within one pound, then all that we can say is that his weight without his cat is really somewhere between 185.5 and 186.5 lbs and his weight (displayed value 193 lbs) with his cat is somewhere between 190.5 and 191.5 lbs. The difference between the two is somewhere between 4 and 6 lbs (190.5-186.5 = 4; 191.5-185.5=6). This means that his estimate of 5 lbs can be out by as much as 20% above or below the cat’s actual weight.

I don’t know the details of the electronics in my scales but I have noticed that certain digital thermometers appear to be designed around 1024 bit ADCs and are scaled such that one bit is identically equivalent to 0.1 °C giving a range of -50 °C  and +50 °C.  Conversion to Fahrenheit is done in firmware rather than the hardware as this simplifies the hardware production but with the cost of adding a possible small additional source of inaccuracy to the instrument due to digital rounding. I assume that this is the same in bathroom scales.

In short, the original writer should be weighing his cat on a proper scale which is designed for the job, not on a bathroom scale. Our dog knows that whenever he goes to the vets, if he jumps onto the vet’s scales (which are of course in kilograms) he gets a treat.