(CNN) -- Forget the needle. Forget the haystack.
Searchers in the south Indian Ocean are narrowing in on a series of pings — weakening 33.3 kHz pings -- that they heard twice over the weekend and two times again Tuesday.
Here's how they are doing it:
As you know by now, the Australian vessel Ocean Shield began the underwater phase of the search late last week, lowering the familiar wing-shaped towed pinger locator (TPL-25) into the sea.
The TPL is connected to the ship by about 4,500 meters (2.8 miles) of cable. The cable is so long that, were the TPL allowed to sink, it might actually touch the deep ocean floor. But as the ship sails, the TPL is pulled at a depth of about 3,000 meters (about 1.9 miles), some 1,000 to 1,500 meters (0.6 to 0.9 miles) above the bottom.
From that position, the TPL can detect pings roughly one mile in any direction.
Call it a triumph of science, or incredible luck, but on the very first path, the Ocean Shield, which was following a path suggested by an analysis of Inmarsat satellite data, detected a steady series of pings Saturday afternoon, Perth, Australia time.
As the ship sailed at a leisurely 2 knots (roughly 2.3 miles per hour), the ship continued hearing the pings, monitoring the electronic pulse for 2 hours and 20 minutes as it got stronger, and then, over time, weaker.
Technicians recorded the signals on computers and analyzed their strength, marking the TPL's location at the spot where it detected the peak amplitude, or maximum strength, said Michael Dean, the U.S. Navy's deputy director of ocean engineering.
Two pings and then ....
Once the first "run" was completed, the ship began the time-consuming job of turning around. Because the ship is towing the pinger locator, it can take many hours to turn the vessel around, even when they reel in the TPL to allow a tighter turn. Australian search coordinator Angus Houston said the turning around takes three hours; Dean said it can take as many as eight.
The ship then began a parallel run Saturday night Perth time about 3/4th of a mile away from the original run, close enough that the TPL's ability to detect pings overlaps with the previous pass. The overlap is as much as 50%, Dean said.
Again, the technicians record signals, taking note of the point where the pings are strongest.
In the case of the Ocean Shield, technicians heard two pings on the second run, but for only 13 minutes.
That only increased the excitement, because Malaysia Airlines Flight 370 has two pingers -- one on the flight data recorder and one on the cockpit voice recorder.
After conducting the two runs, searchers conducted more runs adjacent the earlier sweeps, but they did not detect any pings. That led authorities to express concern that perhaps the batteries had weakened or even died.
But the "runs" continued, and Tuesday, there were two more encounters with the pings. In one run, pings were detected for about 5 minutes and 32 seconds; on the other, about 7 minutes.
Authorities did not say whether the sequence of the runs — which side of the original run they were on. But, Dean said, in traditional searches, runs are made on either side until they are confident they have "boxed in" the strongest signal.
They then begin a series of runs perpendicular to the first set of runs, again trying to "box in" the strongest single.
"Essentially, when you are done you're going to have a tic-tac-toe board and you'll have boxed it in both directions," Dean said.
The box with the strongest signal is where you search.
Detected a different frequency
"I believe we are searching in the right area, but we need to visually identify wreckage before we can confirm that this is the final resting place of MH370," said a cautious Houston. "This is absolutely imperative."
'Hopefully, with lots of (pinger) transmissions, we will have a tight small area" to search, Houston said, saying they may be able to find the wreckage "in a matter of days."
But the TPL appears to have arrived in the nick of time. The fact that pings are being heard for decreasing lengths of time may be due to the batteries dying, Houston said. The fact that only one of the two pingers has been heard during the most recent runs may be evidence that one pinger battery has already died, he said.
Experts said they were not concerned that the pings were detected at a frequency of 33.331 kHz, instead of the design frequency of 37.5.
"We're listening a little bit on either side of that (37.5 kHz) because pinger (frequencies) do drift," Dean said.
The pingers did have the same pulse rate as the MH370 pingers — one ping per second. Searchers "believe the signals to be consistent with the specification and description of a flight data recorder," Houston said.
By using the TPL, searches can greatly reduce the time it takes to locate the missing plane. The TPL can cover terrain six times faster than the Bluefin-21 AUV, the autonomous underwater vehicle that will eventually be deployed to scan and photograph the ocean floor.
If the pingers keep working, the search area can be narrowed to within "a couple of hundred meters of the target," Dean said.
"At this point, as tempting as it is to launch an AUV, you really want to remain rigorous in running lines and seeing if you can reacquire a signal," he said.