On October 15, an investigative report by The Wall Street Journal said that hot startup Theranos, founded by Elizabeth Holmes, was mostly relying on traditional technology to analyze blood-test results.

Later, on CNBC's "Mad Money," Holmes said that as part of a voluntary transition to getting its tests approved by the US Food and Drug Administration, the company stopped using its proprietary "nanotainers" to collect finger pricks of blood — though some reports indicate that pressure from the FDA was what fueled that decision.

Many of these reports have observers wondering whether Theranos' supposedly revolutionary blood tests work at all.

Until either Theranos shows the data that Holmes has said disproves allegations from former employees that its tests don't work as well — we've requested that data multiple times — or until the FDA releases its decisions on the other approximately 130 tests the company has submitted for review, we won't know the answer to that question.

What we can do is look at existing science and expert opinion to try to make an educated guess about how these blood tests are supposed to work.

In April, Tech Insider talked to experts in the fields of clinical pathology and laboratory medicine, microfluidics, and biomedical engineering to try to piece together an answer.

No one could figure it out conclusively because, as Dr. David Koch, president of the American Association for Clinical Chemistry and a professor at Emory University, said then, "It may be wonderful [or] it may bomb, but I really can't be more definitive because there's nothing to really look at, to read, to react to."

Without being able to see how Theranos' technology works, experts still can't say whether or not it does work — but a description of the process, which Theranos recently described briefly itself, is still possible, and we've learned some new things since April.

Here's a step-by-step breakdown.

1. The blood draw

Here's the first place Theranos — in theory — sets itself apart from competitors. Part of the promise that attracted all of the initial attention to Theranos was that you could get your blood tested without a "big, scary needle" with blood drawn from just a finger prick.

In theory, that blood was then collected in a proprietary piece of technology known as a nanotainer, a tiny container barely half an inch high, just big enough to hold a couple of drops.

In April, a representative for Theranos told Tech Insider: "We can perform hundreds of tests, from standard to sophisticated, from a pinprick and tiny sample of blood, and we have performed more than 70 tests from a single tiny sample."

For now, Theranos is only using those nanotainers commercially for the one test Theranos has received FDA approval for, a test for the herpes simplex 1 virus.

Until FDA approval is granted for additional tests, Theranos is conducting the majority of its tests using traditional venous blood draws, with a needle in the arm. They could still be doing some finger prick testing without the nanotainer, but the company has not been willing to say how frequently they are doing so.

2. Transportation to a lab

Theranos currently has two labs where it can process the samples drawn at its consumer-facing locations. One "high-complexity" lab in Newark, California, has Theranos' proprietary "Edison" blood analysis technology — along with other traditional machines — while the other "medium-complexity" lab in Scottsdale, Arizona, has only traditional technology.

theranos blood testTheranos/Screenshot

3. Analysis

Here's where stuff gets tricky. Each test is different, Dr. Jerry Yeo, a professor and director of Clinical Chemistry Laboratories at the University of Chicago, tells Tech Insider. And while he says that he can't specifically comment on how good Theranos' tests are, "every test has a limitation," he says.

Processing a tiny amount of fluid and getting an accurate result is what Theranos needs to be able to do if its technology really works.

"Every test that we offer in our lab can be run on our proprietary devices," Holmes told Cramer on "Mad Money," meaning that the Edison machines can run all of the tests they've submitted to the FDA.

The Wall Street Journal story casts doubt on that claim, with allegations from anonymous former employees saying that these tests were not as accurate as Theranos says, especially tests for Vitamin D, prostate cancer, and thyroid hormones. While Holmes has said the company has proof that its tests are as accurate as traditional ones, we're still waiting to see it.

The specifics of how those Edison machines work is a mystery, and in reality, the process would be different for each test that Theranos offers. The technology that provides the ability to work with tiny flows of fluid is called microfluidics, but each test would require a different process.

Yeo tells TI that a herpes test could potentially extract DNA from a tiny bit of blood and then multiply it using a technology called PCR to provide enough information about the herpes virus. But we don't know if that's exactly what Theranos' process is, and even if so, that doesn't mean all other tests would work the same way.

It's possible that the Edison machines are accurate with some tests and are more likely to be wrong with others — we just don't know.

That uncertainty — and the need to process all kinds of blood samples quickly — may be why Theranos is relying more on traditional blood analysis machines made by companies like Siemens for now.

In some cases, a tiny sample may provide enough material for these machines to run a specific test. In other cases, the sample may need to be diluted.

Dilution is sometimes necessary, for example, when not enough blood is drawn to run a test. But it raises the risk of error, so it's not usually part of regular practice.

theranos chairmanSteve Jennings/Getty Images

Heather King, Theranos' general counsel, emailed Fortune's Roger Parloff and said that Theranos doesn't dilute samples "in the way that the Journal described." But dilution is a complex process, and while it can be used to run different tests, Yeo explains that there are many different ways of diluting a sample, using automated processes and different dilution liquids.

Even if they don't dilute samples "in the way" the Journal described, that doesn't answer the question of whether they dilute samples or not, something that Yeo and others have described as having the potential to lead to errors. The Journal quoted one lab expert who said regularly that diluting samples was "poor laboratory practice."

And of course, if the sample is the amount of blood drawn from a traditional blood draw and processed in a normal machine made by a third party like Siemens, the question is moot — we already know that those systems work.