We need to start at the beginning: how did Hansen and his team construct this figure?

At the right of the figure, from 1870 to 2005, modern measurement temperatures of the upper layer of the ocean are plotted. This is called the Sea Surface Temperature (SST) . Although there are several techniques to determine the SST, nowadays the most commonly used measurements come from satellite data. Before the satellite age, people used buoys with attached thermometers and, even more simply, pulling a bucket of sea water and sticking a thermometer in it. There's also some wriggle room left how to connect this patchwork of datasets, but let's not go there. The whole of it gives a reasonable reliable record.

At the left of the vertical line, another set of data has been plotted. Although plotted as temperature, they are not direct measurements, but inferred from a proxy. And in this case: the shell of a particular critter of the ocean is used: the foraminifer G. ruber (see picture right). Foraminifera are miracles of evolutionary creativity worthy of its own diary of wonders. They are to this day able to completely frustrate earth scientists and biogeologists. Foraminifera are single celled organism which are capable of forming a calcium carbonate shells (or sometimes even from silica). Without this capacity of locking carbon away into gigantic limestone formations (think: Dover, Brittany) there wouldn't have been ice ages and the world today would've been a lot hotter - but the important bit here is that their shells are used for establishing a paleo-temperature record.

How does that work? Synopsised: People take a core of the ocean's sediment, sample it in tiny bits, filter it for specific foraminifera (which means: picking them by hand), crush the shells and run the powder through a spectrometer to establish the concentration of two elements in particular: Calcium (Ca) and Magnesium (Mg). Magnesium is an element which can substitute for calcium in the calcium carbonate shell but - and this is the clincher - the concentration of magnesium in the shells varies under certain conditions. One of them being: water temperature.

Therefore, if people take a sediment core, sample various ages, determine the Mg/Ca ratio in the specific foraminifera and plot it, then bingo: you've a technique that is indicative of a temperature of that time. Or with the jargon: you have a paleo-thermometer.

Hansen et al took the Mg/Ca data from a sediment core which was taken nearby the SST temperature measurements and then combined the two into one graph, a process which has been coined splicing by those critical of that approach. Because what happens is that two different kinds of datasets are juxtaposed and presumed equal.

And here start the problems with the whole figure: that's still a pretty big presumption, and Hansen's article spends barely a few lines on the diverse and complex difficulties still present with establishing a reliable temperature proxy.

It does great injustice to a whole branch of science in foraminifera proxies by wrapping it up with this discussion: "The paleoclimate SST, based on Mg content of foraminifera shells, provides accuracy to ~1°C." And a few lines more.

Because foraminifera proxies do not equal the SST. The G. rubber foraminifer floats relatively shallow beneath the ocean surface - but it's not at the surface. So a conversion is needed. This is supposed to have been done already - but it's not mentioned. And articles on this topic sprinkle with various confidence intervals, from 0.6 to 1.7°C.

Secondly, a more serious problem which is currently deeply under research in this field arises: the problem of dissolution after formation. Since when foraminifera die, they sink to the bottom of the ocean and stick around there, until they're retrieved years later with the drilled core. But that implies that they've been up to tens of thousands year exposed to different conditions, which can easily affect the Mg/Ca ratio - and hence affects the paleo-temperature negatively. In that case: the stored Mg/Ca ratio will give you a lower paleo-temeperature.

To this date it has not been decided what a correct conversion factor for that effect would be - or whether it is at all possible to get one. But even a modest attempt would increase the paleo-temperature upward. Combine that with the previous conversion needed and you can end up with a graph that looks like this:

Graph from Climate Audit
Black = Original plot from Hansen et al article
Red = Amended calculation using dissolution corrections

What does this say? To me, it says that, as with many proxy studies, you can't say anything useful at all. Hansen et al use an argument which shouldn't be used as one. With fault ranges this big, the argument becomes invalid - for now.

As long as the accuracy of proxies are hampered by their fault ranges, this kind of criticism is hard to shake off. Foraminifera proxies may be a precise tool, but there is still a set of them that suffers up to this day from being inaccurate. That may change, and I rather see it happening sooner rather than later. It is however a faux pas for climate scientists to ignore these pitfalls- it is not justified by the urgency of the climate debate of today. This way, it won't help to convince people - in fact, I've been growing towards the position that it is counter-productive.

Climate Audit has more critical notes to places - about which I can neither corroborate nor disagree with - such as the alignment of the two datasets, the temperature data itself, and some more. Be that as it may, just turning a blind eye to the problems still surrounding the foraminifera proxies does the science no good.

Climate Audit posts on the Hansen et al article:
Hansen Simplified
Hansen and Bracket Fatigue
The Hansen Splice
Barker 2005 - A good overview of foraminifera as proxies using the Mg/Ca ratio (pdf!!)

And in the Meantime...
Antarctic Sea Ice May Be on the increase (yes, increase!)
Prepare for a New El Nino