My phone rang. It was Adam from Pathology Blawg. I'd sought his help eight months earlier when I was trying to understand the complexities of laboratory billing for one of my stories in the Medicare series. He'd patiently explained to me what lab procedures certain billing codes corresponded to -- knowledge I'd later used to expose a scam at a big operator of cancer treatment centers.

It was the second Monday in February and I was sitting at my messy desk in the Wall Street Journal's Midtown Manhattan newsroom casting about for a new story to sink my teeth into. I'd recently finished work on a year-long investigation of Medicare fraud and had no idea what to do next. After sixteen years at the Journal, this was something I still hadn't mastered: the art of swiftly and efficiently transitioning from one investigative project to the next.

Adam told me he'd stumbled across what he thought could be a big story. People often come to journalists with tips. Nine times out of ten, they don't pan out, but I always took the time to listen. You never knew. Besides, at this particular moment, I was like a dog without a bone. I needed a new bone to chew on.

Now that he mentioned it, there were some things I'd read in that article that I'd found suspect. The lack of any peer-reviewed data to back up the company's scientific claims was one of them. I'd reported about health-care issues for the better part of a decade and couldn't think of any serious advances in medicine that hadn't been subject to peer review. I'd also been struck by a brief description Holmes had given of the way her secret blood-testing devices worked: "A chemistry is performed so that a chemical reaction occurs and generates a signal from the chemical interaction with the sample, which is translated into a result, which is then reviewed by certified laboratory personnel."

Adam asked if I'd read a recent feature in The New Yorker about a Silicon Valley prodigy named Elizabeth Holmes and her company, Theranos. As it turned out, I had. I subscribed to the magazine and often read it on the subway to and from work.

Those sounded like the words of a high school chemistry student, not a sophisticated laboratory scientist. The New Yorker writer had called the description "comically vague."

When I stopped to think about it, I found it hard to believe that a college dropout with just two semesters of chemical engineering courses under her belt had pioneered cutting-edge new science. Sure, Mark Zuckerberg had learned to code on his father's computer when he was ten, but medicine was different: it wasn't something you could teach yourself in the basement of your house. You needed years of formal training and decades of research to add value. There was a reason many Nobel laureates in medicine were in their sixties when their achievements were recognized.

Adam said that he'd had a similar reaction to the New Yorker piece and that a group of people had contacted him after he'd posted a skeptical item on his blog about it. He was cryptic about their identities and their connection to Theranos at first, but he said they had information about the company I'd want to hear. He said he'd check with them to see if they were willing to talk to me.

In the meantime, I did some preliminary research on Theranos and came across the Journal's editorial-page piece from seventeen months earlier. I hadn't seen it when it was published. This added an interesting wrinkle, I thought: my newspaper had played a role in Holmes's meteoric rise by being the first mainstream media organization to publicize her supposed achievements. It made for an awkward situation, but I wasn't too worried about it. There was a firewall between the Journal's editorial and newsroom staffs. If it turned out that I found some skeletons in Holmes's closet, it wouldn't be the first time the two sides of the paper had contradicted each other.

Two weeks after our initial conversation, Adam put me in touch with Richard and Joe Fuisz, Phyllis Gardner, and Rochelle Gibbons. It was disappointing at first to hear that the Fuiszes had been involved in litigation with Theranos. Even if they insisted they'd been wrongly accused, the lawsuit gave them a big ax to grind and made them useless as sources.

But my ears pricked up when I heard that they had talked to Theranos's just-departed laboratory director and that he was alleging some sort of wrongdoing at the company. I also found the story of Ian Gibbons tragic and was intrigued by the fact that Rochelle said he'd confided to her on several occasions that the Theranos technology wasn't working. It was the type of thing that would have been dismissed as hearsay in court, but it seemed credible enough to merit a closer look. In order to take this any further, though, what I needed to do next was clear: I needed to talk to Alan Beam.

The first half dozen times I dialed Alan's number, I got his voicemail. I didn't leave a message and instead resolved to just keep trying him. On the afternoon of Thursday, February 26, 2015, a voice with an accent I couldn't quite place finally answered the phone. After ascertaining that it was in fact Alan, I introduced myself and told him I understood he had recently left Theranos with concerns about the way the company was operating.

It took me a while to understand the dilution part. Why would they do that and why was it bad? I asked. Alan explained that it was to make up for the fact that the Edison could only do a category of tests known as immunoassays. Theranos didn't want people to know its technology was limited, so it had contrived a way of running small finger-stick samples on conventional machines. This involved diluting the finger-stick samples to make them bigger. The problem, he said, was that when you diluted the samples, you lowered the concentration of analytes in the blood to a level the conventional machines could no longer measure accurately.

With the ground rules for our conversation established, Alan let down his guard and we talked for more than an hour. One of the first things he said was that what Ian had told Rochelle was true: the Theranos devices didn't work. They were called Edisons, he said, and were error prone. They constantly failed quality control. Furthermore, Theranos used them for only a small number of tests. It performed most of its tests on commercially available instruments and diluted the blood samples.

I could sense he was very nervous, but he also seemed to want to unburden himself. He told me he would speak to me only if I promised to keep his identity confidential. Theranos's lawyers had been harassing him and he was certain the company would sue him if it found out he was talking to a reporter. I agreed to grant him anonymity. It wasn't a hard decision. Without him, all I had were secondhand sources and informed speculation. If he wouldn't talk, there would be no story.

He said he had tried to delay the launch of Theranos's blood tests in Walgreens stores and had warned Holmes that the lab's sodium and potassium results were completely unreliable. According to Theranos's tests, perfectly healthy patients had levels of potassium in their blood that were off the charts. He used the word "crazy" to describe the results. I was barely getting my head around these revelations when Alan mentioned something called proficiency testing. He was adamant that Theranos was breaking federal proficiency-testing rules. He even referred me to the relevant section of the Code of Federal Regulations: 42 CFR, part 493. I wrote it down in my notebook and told myself to look it up later.

Alan also said that Holmes was evangelical about revolutionizing blood testing but that her knowledge base in science and medicine was poor, confirming my instincts. He said she wasn't the one running Theranos day-to-day. A man named Sunny Balwani was. Alan didn't mince his words about Balwani: he was a dishonest bully who managed through intimidation. Then he dropped another bombshell: Holmes and Balwani were romantically involved. I knew from reading the New Yorker and Fortune articles and from browsing the Theranos website that Balwani was the company's president and chief operating officer. If what Alan was saying was true, this added a new twist: Silicon Valley's first female billionaire tech founder was sleeping with her number-two executive, who was nearly twenty years her senior.

Alan said he had raised his concerns about proficiency testing and the reliability of Theranos's test results with Holmes and Balwani a number of times in person and by email. But Balwani would always either rebuff him or put him off, making sure to copy a Theranos lawyer on their email exchanges and to write, "Consider this attorney-client confidential."

As the laboratory director whose name had been on the Theranos lab's CLIA license, Alan was worried that he would be held personally responsible if there was ever a government investigation. To protect himself, he told me he'd forwarded dozens of his email exchanges with Balwani to his personal email account. But Theranos had found that out and threatened to sue him for breaching his confidentiality agreement.

It was sloppy corporate governance, but then again this was a private company. There were no rules against that sort of thing in Silicon Valley's private startup world. What I found more interesting was the fact that Holmes seemed to be hiding the relationship from her board. Why else would the New Yorker article have portrayed her as single, with Henry Kissinger telling the magazine that he and his wife had tried to fix her up on dates? If Holmes wasn't forthright with her board about her relationship with Balwani, then what else might she be keeping from it?

What worried him even more than any personal liability he might face was the potential harm patients were being exposed to. He described the two nightmare scenarios false blood-test results could lead to. A false positive might cause a patient to have an unnecessary medical procedure. But a false negative was worse: a patient with a serious condition that went undiagnosed could die.

I hung up the phone feeling the familiar rush I got whenever I made a big reporting breakthrough and had to remind myself that this was just the first step in a long process. There was still a lot to understand and, above all, the story would require corroboration. There was no way the paper would take it with just one anonymous source, however good that source might be.

The next time Alan and I talked, I was standing in Brooklyn's Prospect Park trying to stay warm while keeping a loose eye on my two boys, ages nine and eleven, as they horsed around with one of their friends. It was the last Saturday in what would go down in the record books as New York City's coldest February in eighty-one years.

I told Alan that I felt like I was beginning to make progress, which he seemed pleased to hear. I asked him whether he had kept the emails he'd forwarded to his personal Gmail account. My heart sank when he responded that his lawyer had made him delete them to comply with the affidavit the company made him sign. Documentary evidence was the gold standard for these types of stories. This would make my job much more difficult. I tried not to betray my disappointment.

I had texted Alan after our first conversation to ask if he could think of former colleagues who might corroborate what he'd told me. He'd sent seven names, and I'd made contact with two of them. Both had been extremely nervous and had only agreed to talk on deep background. One of them, a former Theranos CLS, wouldn't say much, but what she did say gave me confidence that I was on the right track: she told me she had been very troubled by what was going on at the company and concerned for patient safety. She'd resigned because she wasn't comfortable having her name continue to appear on test results. The other was a former technical supervisor in the lab who'd said that Theranos operated under a culture of secrecy and fear.

He also mentioned that Theranos was working on a newer-generation device code-named 4S that was supposed to supplant the Edison and do a broader variety of blood tests, but it didn't work at all and was never deployed in the lab. Diluting finger-stick samples and running them on Siemens machines was supposed to be a temporary solution, but it had become a permanent one because the 4S had turned into a fiasco.

Our conversation shifted to proficiency testing. Alan explained how Theranos was gaming it and he told me which commercial analyzers it used for the majority of its blood tests. Both were made by Siemens, confirming what Andrew Perlman, Phyllis Gardner's husband, had heard from a Siemens sales representative during a flight. He revealed something else that hadn't come up in our first call: Theranos's lab was divided into two parts. One contained the commercial analyzers and the other the Edison devices. During her inspection of the lab, a state inspector had been shown only the part with the commercial analyzers. Alan felt she'd been deceived.

It was all beginning to make sense: Holmes and her company had overpromised and then cut corners when they couldn't deliver. It was one thing to do that with software or a smartphone app, but doing it with a medical product that people relied on to make important health decisions was unconscionable. Toward the end of this second phone conversation, Alan mentioned something else I found of interest: George Shultz, the former secretary of state who was a Theranos board member, had a grandson named Tyler who had worked at the company. Alan wasn't sure why Tyler had left but he didn't think it was on good terms. I was jotting things down in the Notes app of my iPhone and added Tyler's name as another potential source.

Over the next few weeks, I made some more progress but I also encountered some complications. In my quest to corroborate what Alan was telling me, I contacted more than twenty current and former Theranos employees. Many didn't reply to my calls and emails. The few that I managed to get on the phone told me they had signed ironclad confidentiality agreements and didn't want to risk being sued for violating them.

A few days later, Alan got back in touch with some good news. I had asked him to call the Washington, D. C., whistleblower law firm he'd reached out to in the fall to see if he could retrieve the email exchange with Balwani he had sent to it. The firm had just complied with his request. Alan forwarded the exchange to me. It was a chain of eighteen emails about proficiency testing between Sunny Balwani, Daniel Young, Mark Pandori, and Alan. It showed Balwani angrily admonishing Alan and Mark Pandori for running the proficiency-testing samples on the Edison and reluctantly acknowledging that the device had "failed" the test. Moreover, it left no doubt that Holmes knew about the incident: she was copied on most of the emails.

One former high-ranking lab employee did agree to talk to me but only off the record. This was an important journalistic distinction: Alan and the other two former employees had agreed to speak to me on deep background, which meant I could use what they told me while keeping their identities confidential. Off the record meant I couldn't make any use of the information. The conversation was nonetheless helpful because this source confirmed a lot of what Alan had told me, giving me the confidence to forge on. He summed up what was going on at the company with an analogy: "The way Theranos is operating is like trying to build a bus while you're driving the bus. Someone is going to get killed."

Although it was a big setback, I was slowly making headway on other fronts. Wanting a neutral opinion from a lab expert about Theranos's dilution of blood samples and the way it conducted proficiency testing, I called Timothy Hamill, vice chairman of the University of California, San Francisco's Department of Laboratory Medicine. Tim confirmed to me that both practices were highly questionable. He also explained the pitfalls of using blood pricked from a finger. Unlike venous blood drawn from the arm, capillary blood was polluted by fluids from tissues and cells that interfered with tests and made measurements less accurate. "I'd be less surprised if they told us they were time travelers who came back from the twenty-seventh century than if they told us they cracked that nut," he said.

This was another step forward, but it was soon followed by a step backward. In late March, Alan got cold feet. He stood by everything he'd told me, but he no longer wanted to be involved with the story going forward. He couldn't stomach the risks anymore. Talking to me gave him palpitations and distracted him from his new job, he said. I tried to get him to change his mind but he was resolute, so I decided to give him some space and hoped he would eventually come around.

Before his change of heart, Alan had mentioned a nurse in Arizona named Carmen Washington who worked at a clinic owned by Walgreens and had complained about Theranos's blood tests. After trying to track her down for several weeks, I finally got her on the phone. She told me three of her patients had received questionable results from the company. One was a sixteen-year-old girl with a sky-high potassium result that suggested she was at risk of a heart attack. The result hadn't made sense given that she was a teenager and in good health, Carmen said. Two other patients had received results showing abnormally high levels of thyroid-stimulating hormone, or TSH. Carmen had called them back in to the clinic and redrawn their blood. The second time, their results had come back abnormally low. After that, Carmen had lost faith in Theranos's finger-stick tests. These incidents tracked with Alan's claims. TSH was one of the immunoassays Theranos performed on the Edison that had failed proficiency testing.

Carmen Washington's story was helpful, but I soon had something better: another Theranos whistleblower. I had dropped Tyler Shultz a note through LinkedIn's InMail messaging feature after noticing that he had viewed my profile on the site. I figured he must have heard from other former employees that I was poking around. It had been more than a month since I'd made the overture and I was losing hope that he would reply when my phone rang.

"He made it through Watergate and the Iran-Contra scandal with his integrity intact," Tyler told me. "I'm sure he'll get through Theranos if he's still alive to make things right."

Tyler's motivation for talking to me was twofold. Like Alan, he was worried about patients getting inaccurate test results. He was also concerned for his grandfather's reputation. Although he felt certain Theranos would eventually be exposed, he wanted to hasten the process to give his grandfather the chance to clear his name. George Shultz was ninety-four and might not be around all that much longer.

It was Tyler and he seemed eager to talk. However, he was extremely worried that Theranos would come after him. He was calling me from a burner phone that couldn't be traced back to him. After I agreed to grant him confidentiality, he told me in broad strokes the story of his eight months at the company.

On his way out the door, Tyler had printed his email to Holmes and Balwani's response and smuggled them out under his shirt. He also still had the emails he'd exchanged with the New York State Health Department about proficiency testing. This was music to my ears. I asked him to send me everything, which he promptly did.

It was time to head to Palo Alto. But before going, there was somewhere else I wanted to visit first.

I had another lead, though, after scanning Yelp to see if anyone had complained about a bad experience with Theranos. Sure enough, a woman who appeared to be a doctor and went by "Natalie M." had. Yelp has a feature that allows you to send messages to reviewers, so I sent her a note with my contact information. She called me the next day. Natalie M.'s real name was Nicole Sundene. She was a family practitioner in the Phoenix suburb of Fountain Hills and she was very unhappy with Theranos. The previous fall, she had sent one of her patients to the emergency room because of a frightening lab report from the company only to find out it was a false alarm. I flew to Phoenix to meet Dr. Sundene and her patient. While there, I also planned to drop in unannounced on other physician practices that used Theranos for their lab tests. I'd gotten the names of a half dozen from an industry source.

I needed to prove that the company was producing inaccurate blood-test results. The only way to do that was to find doctors who had received questionable lab reports and sent their patients to get retested elsewhere. The best place to go looking was Phoenix, where Theranos had expanded to more than forty locations. My first thought had been to pay a visit to Carmen Washington, but she'd left the Walgreens clinic she worked at on the corner of Osborn Road and Central Avenue and didn't have the names of the three patients she'd told me about.

Glunz's case was compelling because it showed both the emotional and the financial toll of a health scare brought on by inaccurate results. As an independent real estate broker, she was self-insured and had a health plan with a high deductible. The ER visit and subsequent MRIs had cost three thousand dollars -- a sum she'd had to pay out of her own pocket.

Dr. Sundene's patient, Maureen Glunz, agreed to meet at a Starbucks near her home. A petite woman in her mid-fifties, she was Exhibit A for one of the two scenarios Alan Beam worried about. The lab report she'd received from Theranos had shown abnormally elevated results for calcium, protein, glucose, and three liver enzymes. Since she had complained of ringing in her ear (later determined to be caused by lack of sleep), Dr. Sundene had worried she might be on the cusp of a stroke and sent her straight to the hospital. Glunz had spent four hours in the emergency room on the eve of Thanksgiving while doctors ran a battery of tests on her, including a CT scan. She'd been discharged after a new set of blood tests performed by the hospital's lab came back normal. That hadn't been the end of it, however. As a precaution, she'd undergone two MRIs during the ensuing week. She said she'd finally stopped worrying when those had come back normal too.

With Dr. Sundene's help, I decided to conduct a little experiment. She wrote me a lab order and I took it to the Walgreens closest to my hotel the next morning, making sure to fast to ensure the most accurate readings. The Theranos wellness center inside the Walgreens wasn't much to behold: it was a little room not much bigger than a closet with a chair and little bottles of water. Unlike Safeway, the pharmacy chain hadn't spent a fortune remodeling its stores to create upscale clinics. I sat down and waited for a few minutes while the phlebotomist entered my order into a computer and talked to someone on the phone. After she hung up, she asked me to lift up my shirtsleeve and wrapped a tourniquet around my arm. Why no finger stick? I asked. She replied that some of the tests in my order required a venous draw.

When I met with Dr. Sundene at her office, I learned that Glunz wasn't the only patient whose results she'd found suspect. She told me more than a dozen of her patients had tested suspiciously high for potassium and calcium and she doubted the accuracy of those results as well. She had written Theranos a letter to complain but the company hadn't even acknowledged it.

I spent the next few days knocking on the doors of other doctors' offices. At a practice in Scottsdale, I talked to Drs. Adrienne Stewart, Lauren Beardsley, and Saman Rezaie. Dr. Stewart described a patient of hers who had postponed a long-planned trip to Ireland at the last minute because of a test result from Theranos suggesting she might have deep vein thrombosis, a condition that occurs when a blood clot forms, usually in the legs. People with DVT aren't supposed to fly because of the risk the clot will break loose, travel through the bloodstream, and lodge in the lungs, causing a pulmonary embolism. Dr. Stewart had subsequently set aside the Theranos result when ultrasounds of the patient's legs and a second set of blood-test results from another lab had been normal.

I wasn't entirely surprised. Alan Beam had explained to me that, of the more than 240 tests Theranos offered on its menu, only about 80 were performed on small finger-stick samples (a dozen on the Edison and another 60 or 70 on the hacked Siemens machines). The rest, he'd said, required what Holmes had likened in media interviews to a medieval torture mechanism: the dreaded hypodermic needle. I now had my confirmation of that. After walking out of the Walgreens, I drove my rental car to a nearby LabCorp site and submitted to another blood draw. Dr. Sundene promised to send me both sets of results when they arrived. Come to think of it, she would get herself tested at both places too, to broaden our comparative sample, she said.

I also met with Dr. Gary Betz, a family practitioner in another part of town who had stopped sending his patients to Theranos after a bad experience involving one of them the previous summer. That patient, also a woman, was on medication to reduce her blood pressure. One of the medicine's potential side effects was high potassium, so Dr. Betz monitored her blood regularly. After Theranos reported a near critical potassium value for the patient, a nurse in Dr. Betz's office sent her back there to get tested again to make sure the result was correct. But during the second visit, the phlebotomist made three unsuccessful attempts to draw her blood and then sent her home. Dr. Betz was furious when he found out the next day: if the original result was correct, it was imperative that he get confirmation of it as soon as possible so he could make changes to her treatment. He sent the patient to get retested at Sonora Quest. As it turned out, it was another false alarm: the potassium value Sonora Quest reported that evening was much lower than the Theranos result and well within the normal range. Dr. Betz told me the episode had shattered his trust in Theranos.

The incident had made her leery when Theranos sent her a lab report for another one of her patients showing an abnormally high TSH value. The patient was already on thyroid medication and the result suggested that her dose needed to be raised. Before she did anything, Dr. Stewart sent the patient to get retested at Sonora Quest, a joint venture of Quest and the hospital system Banner Health. The Sonora Quest result came back normal. Had she trusted the Theranos result and increased the patient's medication dosage, the outcome could have been disastrous, Dr. Stewart said. The patient was pregnant. Increasing her dosage would have made her levels of thyroid hormone too high and put her pregnancy at risk.

I wrote Traub back to confirm that I had a story in the works. Could he arrange an interview with Holmes and a visit of Theranos's headquarters and laboratory? I asked. I told him I planned on traveling to the San Francisco Bay Area at the beginning of May, which was about two weeks away, and could meet with her then. He said he would check Holmes's schedule and get back to me.

As I was wrapping up my trip, I got an email from someone named Matthew Traub. He worked for a public relations firm called DKC and said he represented Theranos. He understood I was working on a story about the company and wanted to know if there was any information he could help me with. The cat was out of the bag, which was just as well. I had planned on contacting the company as soon as I got back to New York. At the Journal, we had a cardinal rule called "No surprises." We never went to press with a story without informing the story subject of every single piece of information we had gathered in our reporting and giving them ample time and opportunity to address and rebut everything.

A few days later, I was back at my desk at the Journal when a mailroom employee handed me a thick envelope. It was from Dr. Sundene. Inside were our lab reports from Theranos and LabCorp. As I scanned my results, I noticed a number of discrepancies. Theranos had flagged three of my values as abnormally high and one as abnormally low. Yet on LabCorp's report, all four of those values showed up as normal. Meanwhile, LabCorp had flagged both my total cholesterol and LDL cholesterol (otherwise known as bad cholesterol) as high, while the Theranos report described the first as "desirable" and the second as "near optimal."

Those differences were mild compared to a whopper Dr. Sundene had found in her results. According to Theranos, the amount of cortisol in her blood was less than 1 microgram per deciliter. A value that low was usually associated with Addison's disease, a dangerous condition characterized by extreme fatigue and low blood pressure that could result in death if it went untreated. Her LabCorp report, however, showed a cortisol level of 18.8 micrograms per deciliter, which was within the normal range for healthy patients. Dr. Sundene had no doubt which of the two values was the correct one.

The new source met me at a little brewery called Trappist Provisions on College Avenue in Oakland. She was a young woman by the name of Erika Cheung. Like every other former employee I'd spoken to, Erika was very nervous at first. But as I filled her in about how much information I'd already gathered, she visibly relaxed and began telling me what she knew.

When I heard back from Traub, he told me Holmes's schedule was too booked up to grant me an interview on such short notice. I decided to fly to San Francisco anyway, to meet Tyler Shultz and Rochelle Gibbons in person. There was also another former Theranos employee who was willing to talk to me if I granted her confidentiality.

As someone who had worked in the Theranos lab, Erika had witnessed the December 2013 lab inspection firsthand. Like Alan, she felt the state inspector had been misled. She told me lab members had been under explicit orders not to enter or exit Normandy during the inspection and that the door leading down to it had been kept locked. She also told me about her friendship with Tyler and about the dinner she'd attended at George Shultz's house the night Tyler had resigned. Like Tyler, she was appalled by the lack of scientific rigor that had gone into validating the assays on the Edisons. She said Theranos should never have gone live testing patient samples. The company routinely ignored quality-control failures and test errors and showed a complete disregard for the well-being of patients, she said. In the end, she had resigned because she was sickened by what she had become a party to, she told me. These were strong words, and it was clear from how distraught Erika was that she meant them.

I asked him whether he thought his grandfather was still loyal to Holmes. Yes, there was little doubt in his mind that he was, he replied. When I asked him what made him think that, he revealed a new anecdote. The Shultz family tradition was to celebrate Thanksgiving at the former secretary of state's home. When Tyler, his brother, and his parents had arrived at his grandfather's house that day, they'd come face-to-face with Holmes and her parents. George had invited them too. A mere seven months had passed since Tyler's resignation and the wounds were still fresh, but he had been forced to act as if nothing had happened. The awkward dinner conversation had drifted from California's drought to the bulletproof windows in the new Theranos headquarters. For Tyler, the most excruciating moment had been when Holmes got up and gave a toast expressing her love and appreciation for every member of the Shultz family. He said he'd barely been able to contain himself.

The next day, I drove to Mountain View, home to Google's headquarters, and met Tyler at a beer garden called Steins. It was early evening and the place was packed with young Silicon Valley professionals enjoying happy hour. We couldn't find seats, so we stood around a wooden beer barrel on the terrace outside and used it as a table. Over a pint of cold ale, Tyler gave me a more detailed account of his time at Theranos, including the frantic call from his mother relaying Holmes's threat the day he resigned and his and Erika's attempts to talk sense into George Shultz that evening. He had tried to follow his parents' advice and to put the whole thing behind him but he'd found himself unable to.

I next met up with Phyllis Gardner, the Stanford medical school professor Holmes had consulted about her original patch idea when she'd dropped out of college twelve years earlier. Phyllis gave me a tour of the Stanford campus and its surroundings. As we drove around in her car, I was struck by how small and insular Palo Alto was. Phyllis's home was just down the hill from George Shultz's big shingled house, and both were on land owned by Stanford. When Phyllis walked her dog, she sometimes ran into Channing Robertson. The Hoover Institution building where George Shultz and the other Theranos board members had offices was right in the middle of the campus. The new Theranos headquarters on Page Mill Road was less than two miles away on land that was also owned by Stanford. In a strange twist, Phyllis told me the site used to be a Wall Street Journal printing plant.

Tyler and Erika were both very young and had been junior employees at Theranos, but I found them credible as sources because so much of what they told me corroborated what Alan Beam had said. I was also impressed by their sense of ethics. They felt strongly that what they had witnessed was wrong and were willing to take the risk of speaking to me to right that wrong.

On the last day of my trip, I met Rochelle Gibbons for lunch at Rangoon Ruby, a Burmese restaurant in Palo Alto. It had been two years since Ian had died, but Rochelle was still grieving and struggled to hold back tears. She blamed Theranos for his death and wished he had never worked there. She provided a copy of the doctor's note a Theranos attorney had encouraged Ian to use to avoid being deposed in the Fuisz case. The time stamp on the attorney's email showed that it was sent just a few hours before Ian committed suicide. Rochelle spoke on the record even though she had inherited from her husband Theranos stock options that were potentially worth millions of dollars. She didn't care about the money, she said, and in any case, she didn't believe the shares were really worth anything.

I flew back to New York the next day confident that I'd reached a critical mass in my reporting and that it wouldn't be too long before I could publish. But that was underestimating whom I was up against.