A new device being used on Apple's Cupertino campus running iPhone applications is probably the still-unreleased tablet, according to a metrics company.

Flurry, a San Francisco-based mobile application analytics firm, said that approximately 50 unknown devices at Apple's California headquarters were being tested with a wide variety of iPhone and iPod Touch applications. Flurry tracked the devices using the code that its developer clients embed in their iPhone software to give them usage data.

"We're seeing about 50 [devices] now," said Peter Farago, Flurry's vice president of marketing, in an interview today. "That's up from somewhere in the teens last month."

Flurry is "very confident" that the device it's tracking is the Apple tablet, and not, for instance, a new version of the iPhone, said Farago. "When Apple has new hardware, it's identified as 'iProd', followed by a string of some sort," said Farago, who added that Flurry saw "iProd, 3.1" when Apple was testing the iPhone 3GS last summer.

This device, however, lacked such a string, and was identified to Flurry only as "iProd".

Another clue that it's the tablet being tested is that the devices never leave the Apple campus. All the IP addresses that Flurry detects for the devices are ones assigned to Apple's Cupertino HQ's wireless network, and GPS data pinpoints the location of the hardware to the same area.

"Some of the apps being used happen to have GPS location features," Farago explained.

The fact that the devices don't leave Apple's offices is another good clue that they're tablets being tested by employees. "We've seen iPhones testing [iPhone OS] 4.0 leaving the campus, showing up in an apartment building nearby," said Farago, who added that that's typical of iPhone testing.

"The smoking gun, so to speak, is the specific [iPhone] apps that are being tested," Farago continued. According to Flurry's data, the tablets are running the very things that have been rumoured to be crucial to the tablet's success, with games the top download category, followed by entertainment, then news and books. "We think that these are the kinds of iPhone apps that will be the intended sweet spot for the tablet," Farago continued.

The news and books category has been stressed by analysts, who believe that, at the least, Apple's tablet will take on Amazon's Kindle in the e-reader market. Flurry spotted some of those applications being tested on the tablet, too. "There were a couple [of] dead give-aways of the kinds of media companies that have been rumoured to be part of the launch," said Farago, who declined to name the firms, citing Flurry's agreements with its clients.
Although there's no data that positively, absolutely identifies the device as a tablet - Flurry's API doesn't pass along information on the hardware, such as its screen size, for example-- Farago said the company's engineering team is confident that it has spotted the real deal in use.

"The tablet will run iPhone [and iPod Touch] applications," said Farago, who added it was not clear how they will work on the larger display, whether simply in an expanded mode or as a gadget, occupying only a small section of the tablet screen. "Why would they give up the 130,000 apps in the App Store," he asked rhetorically.

Farago also had an answer for why Apple would arm the tablet with an iPhone OS tagged 3.2 when reports have pegged version 4.0 for a release in late June or early July, at Apple's annual Worldwide Developers Conference (WWDC). The developer's conference is the stage the company's used for iPhone announcements the last two years. "Our assumption is that Apple would like certain publishers to adapt their apps specifically for the tablet, and want them to work on it with enough lead time, so they are having them work on what we assume to be the more stable 3.2," he said. "We think that the new OS for the tablet will be 3.2."

Prognosticators have been divided over what will power the tablet: a variation of the iPhone operating system, Mac OS X, or an entirely new OS. Flurry's data seems to point to the first option of the three.

Flurry first began noticing the assumed tablets in its data last October, but when its researchers dug into older data, they found evidence of a tablet even earlier, said Farago. "Our New York-based team has a history of tracking new Apple hardware pretty accurately," he boasted. "We're the first to have some data to back up some of the rumors about the tablet, and we feel pretty good about our conclusions."

Flurry's tracking code is embedded in about 10,000 iPhone apps, said Farago, and on any given day about 15-20 percent of the apps downloaded from the App Store contain its code.