WEIGHING is the central activity - each of the 3 major markets has a single weighing scale of the spring type, together with entry and exit crushes to enable the passing of a number of cattle. The following problems were observed:
- the scales are repaired when they break down, but they are not calibrated on a periodic basis by a scales servicing agency. At several sites the Project Organisers said that they calibrated the scales from time to time, but their idea of testing consisted typically of testing at 0 kg and/or at 60 kg (a man's weight); this correction (if any) would then be applied to weights in the range of 160 - 400 kg. Thus 3 procedures need to be set up:
- standard weights of whatever kind must be procured or made (sealed
plastic containers partially or completely filled with dry sand may suffice). These weights should make up the values 150, 250, 350 and 400 kg.
- the computer program should incorporate a correction system so that for every market-day the calibration weights and the corresponding scale readings can be entered; this information will be used in the data processing to correct weights and the unit prices.
- under no condition should the weighing recorder manually add or subtract correction quantities to the scale readings before recording information on the data sheets or in the computer (this is being done at Moshi)
- at one of the markets, for a weight of 278 kg, the recorder noted a weight of 268 kg (he accepted and corrected the mistake when I pointed it out)
- the Moshi market has scales which read to 5000 kg; one must assume that they will lose some accuracy if only used in the range 150 - 400 kg..
- in general, all the scales were of the spring type, and require that the
animal stops moving and stays still for long enough for the scale to stabilise; although more expensive, the load-cell type of scales may be preferable, since less time is required per animal to perform the weighing. A load-cell model with a digital readout would also avoid mistakes in scale reading ( as above with the 268 / 278 incident). Typical prices would be in the region of US$ 1250 for a spring model and US$ 2500 for a load-cell model, although this should be verified.
- note that there can be significant changes in weight for animals according to their condition, and according to the availability of water and food on arrival at the market from a trek or rail trip. Typical variations can be +/- 5 kg. This can be particularly a cause of error for Arusha or Moshi when weights are normally taken on the day prior to the market itself.
Note that if there are persistent errors in weighing at a market, it may not really affect reporting for that market itself over time, but it will certainly affect inter-market comparisons.
Possible weighing errors are +/- 1 kg from direct scale reading, +/- 8 kg if scale is incorrect and/or not calibrated, and also +/- 10 kg if the scale type makes reading difficult (e.g. scales calibrated in 2 kg marks, with numeric readings at 20 kg steps.