No one has got vCJD from a blood transfusion yet. And the risk of this happening is low because there have been few cases of the disease so far. Nevertheless, a long-running experiment in which 24 sheep were given blood infected with BSE suggests the disease can indeed spread this way. Two years ago it was revealed that one of the sheep had developed BSE. Now another three have succumbed. Blood transfusion services have already taken steps to prevent the disease spreading in blood from unknowingly infected donors. In earlier tests on rodents, white blood cells emerged as the most infective blood component. So since 1999, Britain's National Blood Service has been filtering out white blood cells from donated blood.

But the latest findings from the sheep experiment -- which will appear next week in the Journal of General Virology -- cast doubt on the effectiveness of this measure. None of the seven animals given only white blood cells have yet developed BSE. The four infected sheep were among the 17 that had whole-blood transfusions.

"From our experiment it's not possible to say where the infectivity is," says Fiona Houston, head of the team at the Institute of Animal Health in Compton, Berkshire. "It could be in the plasma or the red blood cells."

If this is true in humans, red blood cells could provide a refuge for the prion proteins that cause vCJD in humans and BSE in cows. "I think it's important to find out which fractions of blood harbour it," Houston says.

Members of the Britain's independent Spongiform Encephalopathy Advisory Committee agreed when they were told of the results last month. "We should push ahead with fractionated blood to get a handle on where the infective agent is," said committee chairman Peter Smith. "It emphasises even more strongly the potential for transmission in the human situation, and it's telling us that it's a theoretical risk that should be taken very seriously."

But Houston points out that although only 127 people have so far succumbed to vCJD in Britain following the BSE epidemic, that suggests the risk of becoming infected through tainted blood is very low.

In addition, since 1998 the NBS has been buying blood plasma from the US where there have been no cases of BSE or vCJD. Plasma contains valuable components such as clotting agents that can be extracted for treating haemophilia. But because plasma from many people is pooled to provide enough extractable material, a single infected donation could contaminate an entire batch.