Bennett, on the other hand, decided to use a very primitive type of virus that never had the ability to reproduce. Called an adeno-associated virus, these bugs can copy themselves only by borrowing the machinery of the larger, more complex adenovirus.

Even if they were somehow to replicate, it is not a type of virus that causes disease. Of the roughly 300 human gene-therapy trials currently under way in the United States, more than 20 are using these adeno-associated viruses.

On their own, these simple viruses provoke at most a mild response. This is especially true in the eye, an "immuno-privileged" organ in which the immune system does not attack a foreign presence in a significant way.

For her trial, Jean Bennett enlisted the participation of Children's Hospital, a world leader in the treatment and study of pediatric genetic disease.

It is home to the Center for Cellular and Molecular Therapeutics, whose director, prominent gene-therapy specialist Kathy High, agreed to sponsor the trial. A highly specialized lab would purify and tweak the virus so it could perform its job.

A flawed gene

In people with normal vision, a crucial protein is made by the retinal pigment epithelium, a layer of cells beneath the retina. The protein, an enzyme called RPE65, is used to metabolize vitamin A. The metabolized vitamin, in turn, allows the nearby "rod" cells to make a pigment that absorbs light, so it can be converted into an electrochemical message to the brain.

People with an RPE65 mutation develop LCA. Their eyes make flawed versions of the protein, if any at all.

Not only is their vision impaired to begin with, but over the years, their cells accumulate toxic levels of a vitamin A derivative that normally is broken down by RPE65. Their retinas become badly scarred.

To do gene therapy, Bennett's team took the virus and gutted it. They replaced its two genes with a single human gene - one with the blueprint for a healthy version of RPE65. All that remained of the original virus DNA was a pair of genetic bookends, one on each end of the DNA molecule.

Fraser Wright, director of the Children's Hospital lab that made the viral "vector" - the carrier - worked with Jeannette Bennicelli and others to fine-tune it. They added various bits of customized genetic material, including a "promoter" to enhance production of the needed enzyme, and another sequence to make the virus even better at penetrating the patients' eye cells.

The lab grew and purified billions of the viruses. It cultivated them inside cells derived from human kidneys, using a growth medium and other materials that had been subjected to rigorous tests to ensure purity. That process took a whole year.

Wright's lab performed dozens of FDA-mandated tests on the virus itself to ensure it would be safe for people - that there was no contamination by other viruses or bacteria, for instance.

The lab made hundreds of vials and stored them in a freezer.

Nine patients would receive the injection, each time in their worse eye - in case there were complications. The first three were the twins, Tommaso and Josalinda, and another Italian woman named Manuela Migliorati. (The disease is no more common in Italy than anywhere else, but researchers there have made an effort to identify those with mutations.)

If all went well, the viruses would latch onto and enter the sub-retinal cells, then travel all the way to the cells' nuclei. There, each virus would ordinarily release its own DNA. But in the patients, the lab-modified virus instead would release its cargo of the new RPE65 gene.

Because retinal cells do not divide, the impact of the corrective gene would not be watered down over time. It should be permanent.

The first three patients were to get 15 billion viruses each. If everything went OK, the next three would get three times that many. The third group would get 150 billion - 10 times the first dose.

Big numbers, but small invaders.

Compared to the retinal cells that they would be penetrating, each virus was about the size of a Frisbee on a football field.