What is also surprising is how little is known about how the volcano destroyed itself in those fatal last hours.
Most of what we understand is based on a study by Dutch geologist and engineer Roger Verbeek who visited the region shortly after the eruption.
Published in 1885, originally in Dutch, his paper remained relatively obscure for nearly 100 years.
Then in 1983, three years after the catastrophic Mount St Helens lateral blast, two US geologists, Tom Simkin and Richard Fiske, published the first fully comprehensive account of the 1883 eruption that included an English translation of Verbeek's original report.
There is no doubt that the volcanic activity leading up to the killer blasts was of a type known as Plinian (named after Pliny the Younger who witnessed the AD79 eruption of Vesuvius), which sees large quantities of material thrown high into the atmosphere.
But in the words of Simkin and Fiske, "...the critical question of when and how the northern half of Rakata (the largest of the three volcanic cones) collapsed into the sea" remained unanswered.
A granular volcano?
In order to try to shed new light on this, we have been developing a theory, backed up with experiments.
It suggests the volcano may have become dangerously unstable due to pressurisation by gas leaking from the rising magma, helped possibly by steam from boiling seawater that got inside the volcano.
Put simply, the trapped gas pressure helped lower the overall strength of Krakatoa, making it susceptible to massive internal failure.
We speculate that the cause of the huge bang may have been a giant landslide that removed much of the island and exposed magma inside the volcano to the atmosphere.
But for this to happen, the volcano must be made not from solid rock, like Mount Everest, but from material with properties closer to those of sand.
Not all volcanologists agree with the idea of "granular" volcanoes, but our research suggests that it is a good way of thinking about them because it allows us to develop computer models with simple predictive power - a critical step in developing more accurate ways of forecasting future eruptions.
For example, it is widely accepted that pyroclastic rocks entering the sea caused water to be displaced, resulting in the 1883 tsunamis.
But this would also have been helped by underwater slumping of the collapsing volcano, something which our computer models predict.
And by using information on the shape of oceanic volcanoes that have undergone large landslides, for example in Tenerife and Hawaii, combined with estimates from marine surveys on the volumes of slumped material on the sea floor, it might be possible to "back track" the landslide history to reveal the physical conditions needed for this to happen.
Armed with this information, volcanologists might be able to make better judgements about the likely risks of future eruption-generated landslides and accompanying tsunamis.
Professor Nick Petford from Kingston University, London, presents a TV investigation into the science of Krakatoa this Sunday. This follows a feature length drama reconstruction.
Krakatoa, the Last Days: 2100BST BBC One, Sunday 7 May Krakatoa Revealed: 2230BST BBC Two, Sunday 7 May