(Reuters) Coronavirus cases outside China may be ‘tip of the iceberg’: WHO. Quoting,

Much remains to be determined about the virus, which has been linked to a market selling animals in Wuhan, the capital of Hubei province, including how easily it spreads and how deadly it is. Chinese researchers found that its incubation period could be up to 24 days….It had been thought to be up to 14 days. 

The maximum incubation period was thought to be a hard number. There may be no such number, invoking the scenario of the “caged virus.”

The caged virus phenomenon was first written about in conjunction with the rabies virus. Typical incubation periods are weeks, extending to months, occasionally up to two years. Then an isolated case of rabies with an apparent incubation period of 37 years was noted (citation missing.) The concept was further developed with the slow virus.

A slow virus replicates without effective immune response. Some trigger,  perhaps environmental, or possibly analogous to the bacterial quorum sensing, causes a switch from slow to fast.

No “slow” corona viruses have been discovered. There is no reason to think 2019-nCoV is the first one. But the 24 day incubation period, which may have measurement difficulties, suggests a gradation into that.

Most of the things epidemiologists are interested in cannot be derived from “first principles.” Observation is used to select a basic statistical model which can be adjusted. For example,

• A uniform statistical distribution distributes the incubation period  evenly distributed up to a fixed limit, say, 14 days.
• A Poisson distribution has an average incubation period, with tails in the direction of less and more.  It’s typical to cheat by chopping off the “more” tail, perhaps at 14 days, or 24. But this may not be valid.
• We could be looking at a distribution without a practical upper limit for incubation. This doesn’t bother statisticians at all. There are many distributions with infinite variance.

It would mean we have blundered into a virus with a highly effective strategy, with widespread infected carriers, undetectable by any simple, practical means.

In 2019 Novel Coronavirus (2019-nCoV), Wuhan, China; Quarantine is Correct Response, I wrote,

Perhaps by tradition rooted in the 1976 swine flu fiasco, the cost of “no vaccine for 2019-nCoV ” is set for typical epidemiology. Yet the 1918 Spanish flu was untypically deadly for an airborne pathogen….The cost/benefit analysis could have an unpleasant surprise.

This could be that surprise.