“I see. Yes, this is certainly an unusual assignment that you’re on.”

“How many people worked at the carpentry shop?”

“The normal work force was forty or so. I worked there from the age of seventeen in the mid-fifties until the shop closed. Then I became a contractor.” Burman thought for a moment. “This much I can tell you. The guy in your pictures never worked there. He might have been a contractor, but I think I’d recognise him if he was. But there is one other possibility. Maybe his father or some other relative worked at the shop and that’s not his car.”

Mikael nodded. “I realise there are lots of possibilities. Can you suggest anyone I could talk to?”

“Yes,” said Burman, nodding. “Come by tomorrow morning and we’ll go and have a talk with some of the old guys.”

Salander was facing a methodology problem of some significance. She was an expert at digging up information on just about anybody, but her starting point had always been a name and a social security number for a living person. If the individual was listed in a computer file, which everyone inevitably was, then the subject quickly landed in her spider’s web. If the individual owned a computer with an Internet connection, an email address, and maybe even a personal website, which nearly everyone did who came under her special type of research, she could sooner or later find out their innermost secrets.

The work she had agreed to do for Blomkvist was altogether different. This assignment, in simple terms, was to identify four social security numbers based on extremely vague data. In addition, these individuals most likely died several decades ago. So probably they would not be on any computer files.

Blomkvist’s theory, based on the Rebecka Jacobsson case, was that these individuals had fallen victim to a murderer. This meant that they should be found in various unsolved police investigations. There was no clue as to when or where these murders had taken place, except that it had to be before 1966. In terms of research, she was facing a whole new situation.

So, how do I go about this?

She pulled up the Google search engine, and typed in the keywords [Magda] + [murder]. That was the simplest form of research she could do. To her surprise, she made an instant breakthrough in the investigation. Her first hit was the programme listings for TV Värmland in Karlstad, advertising a segment in the series “Värmland Murders” that was broadcast in 1999. After that she found a brief mention in a TV listing in Värmlands Folkblad.

In the series “Värmland Murders” the focus now turns to Magda Lovisa Sjöberg in Ranmoträsk, a gruesome murder mystery that occupied the Karlstad police several decades ago. In April 1960, the 46-year-old farmer’s wife was found murdered in the family’s barn. The reporter Claes Gunnars describes the last hours of her life and the fruitless search for the killer. The murder caused a great stir at the time, and many theories have been presented about who the guilty party was. A young relative will appear on the show to talk about how his life was destroyed when he was accused of the murder. 8:00 p.m.

She found more substantial information in the article “The Lovisa Case Shook the Whole Countryside,” which was published in the magazine Värmlandskultur. All of the magazine’s texts had been loaded on to the Net. Written with obvious glee and in a chatty and titillating tone, the article described how Lovisa Sjöberg’s husband, the lumberjack Holger Sjöberg, had found his wife dead when he came home from work around 5:00. She had been subjected to gross sexual assault, stabbed, and finally murdered with the prongs of a pitchfork. The murder occurred in the family barn, but what aroused the most attention was the fact that the perpetrator, after committing the murder, had tied her up in a kneeling position inside a horse stall.

It was later discovered that one of the animals on the farm, a cow, had suffered a stab wound on the side of its neck.

Initially the husband was suspected of the murder, but he had been in the company of his work colleagues from 6:00 in the morning at a clearing twenty-five miles from his home. It could be verified that Lovisa Sjöberg had been alive as late as 10:00 in the morning, when she had a visit from a woman friend. No-one had seen or heard anything; the farm was five hundred yards from its nearest neighbour.

After dropping the husband as a suspect, the police investigation focused on the victim’s twenty-three-year-old nephew. He had repeatedly fallen foul of the law, was extremely short of cash, and many times had borrowed small sums from his aunt. The nephew’s alibi was significantly weaker, and he was taken into custody for a while, but released for lack of evidence. Even so, many people in the village thought it highly probable that he was guilty.

The police followed another lead. A part of the investigation concerned the search for a pedlar who was seen in the area; there was also a rumour that a group of “thieving gypsies” had carried out a series of raids. Why they should have committed a savage, sexually related murder without stealing anything was never explained.

For a time suspicion was directed at a neighbour in the village, a bachelor who in his youth was suspected of an allegedly homosexual crime-this was back when homosexuality was still a punishable offence-and according to several statements, he had a reputation for being “odd.” Why someone who was supposedly homosexual would commit a sex crime against a woman was not explained either. None of these leads, or any others, led to a charge.

Salander thought there was a clear link to the list in Harriet Vanger’s date book. Leviticus 20:16 said: “If a woman approaches any beast and lies with it, you shall kill the woman and the beast; they shall be put to death, their blood is upon them.” It couldn’t be a coincidence that a farmer’s wife by the name of Magda had been found murdered in a barn, with her body so arranged and tied up in a horse stall.

The question was why had Harriet Vanger written down the name Magda instead of Lovisa, which was apparently the name the victim went under. If her full name had not been printed in the TV listing, Salander would have missed it.

And, of course, the more important question was: was there a link between Rebecka’s murder in 1949, the murder of Magda Lovisa in 1960, and Harriet Vanger’s disappearance in 1966?

On Saturday morning Burman took Blomkvist on an extensive tour of Norsjö. In the morning they called on five former employees who lived within walking distance of Burman’s house. Everyone offered them coffee. All of them studied the photographs and shook their heads.

After a simple lunch at the Burman home, they got in the car for a drive. They visited four villages near Norsjö, where former employees of the carpentry shop lived. At each stop Burman was greeted with warmth, but no-one was able to help them. Blomkvist was beginning to despair.

At 4:00 in the afternoon, Burman parked his car outside a typical red Västerbotten farm near Norsjövallen, just north of Norsjö, and introduced Mikael to Henning Forsman, a retired master carpenter.

“Yes, that’s Assar Brännlund’s lad,” Forsman said as soon as Blomkvist showed him the photographs. Bingo.

“Oh, so that’s Assar’s boy,” Burman said. “Assar was a buyer.”

“How can I find him?”

“The lad? Well, you’ll have to dig. His name was Gunnar, and he worked at the Boliden mine. He died in a blasting accident in the mid-seventies.”

Blomkvist’s heart sank.

“But his wife is still alive. The one in the picture here. Her name is Mildred, and she lives in Bjursele.”

“Bjursele?”