A few simple mistakes in identifying animals and registering their latest performance details can soon add up to a big problem in any recording system. Imagine how much worse the risk of human error becomes when the records relate to a herd of 6000 sows and some of the progeny are selected for future use as breeding stock.

That was the predicament facing the refurbishment of a huge complex about 100 kilometres west of Berlin in Germany to create a breeding-finishing enterprise with its own in-house multiplication programme. Extending to 100 000 square metres of pen area inside buildings that date from the site's past life as a state farm in the former East Germany, the unit at Sandbeiendorf has been developed to produce some 130 000 slaughter pigs per year.

Additionally it is a closed herd for which the replacement gilts are generated on-site and selected Landrace boars are transferred for use in artificial insemination. The only seedstock brought in from outside are some Pietrain boars to use at the farm's separate AI station. Herd replacements for the crossbred Landrace/Large White sows that give the slaughter generation are needed from the pure-line Landrace females that supply the multiplication process. When the unit's owner decided to change his existing recording arrangements, a major concern was therefore to avoid inaccuracies that could result in a stunting of genetic progress because the wrong pigs were chosen or culled.

The vote went to electronic identification as offering greater reliability and accuracy compared to more traditional visual methods. Radio frequency identication (RFID) ear-tags were sourced from Belgian supplier Merko, containing a so-called HiTag S microchip produced by the NXP semiconductors company founded by Philips of the Netherlands. These materials would then be employed to gather data for assembly and analysis through the new Pigs-Online Internet management tool from a subsidiary of Netherlands-based poultry breeder Hendrix Genetics.

Originally devised to allow skilled people on a big unit to run their own breeding programme without importing animals, Pigs-Online has had its first installation at Sandbeiendorf. This involved a back-up of historical data from manually operated PC computer records. Before they could be transferred to the extensive Oracle database being used for the new logistics, however, the records found for approximately 16 000 sows had to be checked that they were accurate.

The team's checks in this instance revealed suspicions about the accuracy of several entries. Ultimately, almost 2000 animals were removed from consideration as candidates to be parents of seedstock because sows lacked the proper pedigree or the information about them was illogical, such as that they had farrowed after fewer than 105 days of gestation. Once checking was complete, the verified records of the remaining sows were used in the calculation of BLUP estimated breeding value (EBV) using a single index combining fertility and grow-finish traits. These values showed a correlation of only 83% with the EBVs obtained previously by the farm's owner, further underlining the vulnerability of genetic improvement programmes to inaccurate recording.

Re-numbering of sows and progeny also needed to be done with great care. The animals had been identified before by means of tattoo numbers and plastic ear-tags. These devices were used to locate the sow on the database so an electronic number could be added for her. The transition demonstrated that about 5% of indicated identities were incorrect, according to the Pigs-Online team supplying the German unit. Sows with a doubtful identity were given a separate numbering code so that they would not interfere with pure-line data. Every member of the breeding herd has since been allocated her new database number on the microchip of her electronic tag and this can be read automatically, which is done at Sandbeiendorf using an Agrident RFID reader integrated into a Psion hand-held computer.

For the herd of 6000 sows it has meant issuing 9 hand-helds to 30 farm workers. Among the operators are the attendants in the mating area, who can now access instant information on identity, estimated breeding value and inbreeding coefficients. Similarly the information on sows for AI becomes visible to the inseminators. The Psion also has a bar-code reading facility. As the pouch for each 100ml semen dose is bar-coded, a completed insemination is registered directly to the database simply by passing the hand-held across the pouch label.

While the supply team was confident about these aspects, however, they confessed afterwards to some concern regarding another feature of the installation. Recorded data were to be relayed automatically through a wireless link to the Internet and that meant equipping the farm with a Wireless Local Area Network (WLAN) system. A survey concluded that 10 access points should be enough, located in 3 of the 7 houses on the site. But they were buildings 400 metres long and divided into compartments. Their ceilings were formed from iron plates, quite apart from the ironwork present in feeding and ventilation equipment that was capable of disrupting wireless signals.

Despite all the initial misgivings, in practice the system has worked well. The access points generate a wireless network throughout the buildings. The hand-held computer readers connect through this network to the Internet and so to the off-site database. Their signal in fact is passed by the nearest access point along a slim optical-fibre cable to the farm office, from which it is relayed through the Internet to the central server running the system's software.

On-line data management has many advantages, the Pigs-Online supply team claims, not least the opportunity for multiple sites of an enterprise or integration to make use of the same database and the possibility for site managers to compare their results with those of other sites. Sandbeiendorf's experience has also demonstrated the value of maintaining herd health through the use of in-house multiplication and the payback possible from automating the recording process because of the labour saved and the accuracy obtained.