Tom Womack

The overnight train takes you from the Former Soviet Union to Mitteleuropa. Cobbled streets rather than eight-lane boulevards, buildings fronted in plaster and pastel paint. Cake shops in vast multitudes, tea and coffee; might as well be Vienna at a sixth the price.

There's a nice hill which looks down onto the historic centre on one side and onto the Soviet side on the other; in fact, several acres of hill with an intriguing maze of unsigned paths around them. There's a certain kind of comfort in going from a city of Orthodox to a city of Catholic churches: I'm much less easy with Orthodox iconography, there's the constant vague fear that you're in the queue to kiss the icon.

The countryside outside Kiev to the north reminds me of nothing more than Thetford Forest, particularly that part just after Mildenhall before you turn off to Go Ape: sandy soil, long straight roads flanked by forestry-commission lattices of fir trees. The road goes ever on and on; there are silver birches planted by the sides of the roads, and then the fields roll, hedgeless, empty of visible labourers and not obviously filled with rows of carefully-weeded crops, to the horizon. The landscape rolls a little - it's not Lincolnshire, if only because there are no huge drainage canals.

After a couple of hours in a minibus, you get to the checkpoint for the First Exclusion Zone; at once the road becomes less well-maintained, and the fields themselves are filled with young silver birches. Twenty more kilometres to the Second Exclusion Zone to pick up a guide and head up to the power plant itself.

Radiation levels are a bit above ambient - three microsieverts an hour at the plant visitor centre, fifteen for a minute or so as the bus drives past a place where it rained at a particularly inopportune time on 26 April 1986. The visitor centre has a fantastic fold-out model of the current state of the plant's interior, as far as it's known - only 60% of the rooms have been explored since the accident, since several of them were hurriedly filled with concrete in the first days of the clean-up without seeing what lay beyond.

What is known, and what the guide described with some terror in her voice, is at least two hundred tons of melted fuel rod mixed with molten rock; at the start of the accident these glowed white hot with nuclear decay, they're now 40 degrees above ambient and the worry is that water can now get in and leach soluble fission products out. The sarcophagus is not watertight; there is a plan to build a huge sealed structure to cover reactor and sarcophagus alike, but the last time I saw that plan the completion date was 2007, and the completion date on the plan on the wall of the visitor's centre is 2012.

The plant sits there under its sarcophagus, square and grey apart from some new yellow steel reinforcing-buttresses which keep the west wall from falling over under the weight of the hastily-installed beams which hold up the hastily-installed roof. The sarcophagus took 206 days of 24-hour work to build, and apparently 90,000 workers: I can't see how they'd fit in, though I imagine all the concrete parts were prefabricated in parallel throughout the Soviet Union, and that an enormous number of workers came in, got their lifetime's permitted radiation dose in a single construction procedure and then left.

There is a never-commissioned fuel storage facility a few kilometres down the road; apparently nobody told the French company building it in the Nineties that Chernobyl's fuel rods were twenty centimetres longer than the standard. You can see the nearly-completed Reactor Five and the start of work at Reactor Six; had all gone to plan and the Soviet Union not fallen, there would be twelve reactors at Chernobyl by now.

Then to Pripyat itself. It's a ghost town, you've seen the documentaries, you've seen the children's play-park rusting away and the secondary school with books discarded on the door, Soviet posters warning you not to swim in unknown waters hanging on the wall, and a copy of a twenty-year-old Pravda browning gently in the corner. It's still pretty striking to see it yourself.

They check you before you leave the Second Exclusion Zone; so for the first time I *know* that my shoes are not radioactive.

Because, consider the alternative:

http://pipeline.corante.com/archives/2008/04/08/fun_with_tunichromes.php#comments

This is, I admit, about as bad as wet-lab chemistry gets; but it's pretty bad. Extracting a vanadium-binding protein, which decomposes in the presence of heat, light, oxygen, or most of the solid supports used in chromatography, from a couple of litres of the blood of sea-squirts, when each sea-squirt supplies very little blood, and your supply of sea-squirts, whilst large, is of indifferent freshness.

I strongly recommend pipeline.corante.com to chemistry groupies; the author is very good at conveying the frustration of medicinal chemistry.

Excel just marked a cell as 'contains possibly wrong formula' because I wasn't adding in the date.

In C1, there is a date; in C2:C10 there are a collection of sums of money, some positive and some negative; in C11 there is sum(C2:C10), and Excel has stuck a little green note in the corner of that cell suggesting that maybe I meant sum(C1:C10).

Dates are stored internally as a count of days since January 1st 1904, so if you look at them as numbers they're around 37000. Adding them to a collection of sums of money which are things like 'income tax deducted from salary this month' or 'pension contribution this month', so of the order of hundreds of pounds, screws up the calculation right royally.

Would anyone be interested in going to a contra dance in Cambridge on Saturday evening? It's at the Emmanuel URC on Trumpington Street, 8-11pm; it's like a ceilidh (so there's a caller, you change partner after each dance, and the music is fiddle and bass), only faster, twirlier and more American.

Along with about half my friends list, I spent the Easter weekend in a hotel near Heathrow airport, at a science fiction convention. Apparently it was the biggest Eastercon for twenty years, but it didn't feel especially crowded.

It was excellent; really without notably annoying flaws. I slept fifteen hours on Friday night, danced wildly until 2am on Saturday, and slept sensibly on Sunday. The rooms had ceilings and decent sound systems, the buffet suppers were mostly Tasty Stew but that's what you want, the infinite cooked breakfasts mean there is now four pounds more fivemack than there was on Thursday. There was lots of space in which you could hear yourself think, there were cheap soft drinks and cheap potatoes so you could keep yourself fed and watered while buying books. A complete success of a con. annafdd was there, which is on its own enough to make any weekend a good weekend. And Mitch Benn, who I'd never previously encountered.

I was on a couple of panels, which seemed to go well, though my observation that, by a lot of common Utopian metrics, life as a citizen of any country in contemporary Western Europe is near enough Utopian did not go down well - does Utopia require that injustice is absent, rather than merely rare, well-reported and noted as bad?

I was not obviously worse prepared or significantly less sure what I was doing than anyone else on the panels, and any panel with the infinitely erudite and impressively composed Edward James on is predestined to reasonably glorious success. I bought nine books, which is much more sensible restraint than the last Eastercon I went to; since I donated a year's worth of do-not-wish-to-reread books to be auctioned for the Alzheimer's Trust, I actually left with lighter bags than I came with.

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19690014329_1969014329.pdf

is a design for a heat exchanger, from the primary sodium-potassium-eutectic coolant loop of a space-based nuclear reactor to the secondary mercury coolant loop. Not quite the chemical fun of a liquid-sodium/water heat exchanger, but I think sodium and potassium both dissolve enthusiastically in mercury to produce amalgams combining the toxicity and gets-everywhere of mercury with the violent reactivity with everything of the alkali metals.

See http://www.dself.dsl.pipex.com/MUSEUM/POWER/mercury/mercury.htm for more examples of the lure of boiling mercury to power-station designers; it seems to be a very appealing improvement on normal superheated steam, at least if you're not too concerned about expense, weight or neurotoxicity.

In a similar vein, http://www.dself.dsl.pipex.com/MUSEUM/POWER/steamwheel/steamwheel.htm talks at length about a variety of designs for steam engines with liquid-metal pistons, dating all the way back to James Watt. You have the choice between mercury, which is not good for kittens and other living things, or bismuth/tin-based alloys, which solidify when the steam-engine is turned off, and sometimes expand enough to burst the engine.

I've made (by accident, I was making yokki and it was supposed to be caramel; I will have to put a dental warning on the shortcakes) a quantity of particularly obdurate toffee. You have to cut it like glass, scratching the surface with a bread-knife and then inserting a sharp knife vertically to cleave it along the scratch; it's prone to shattering.

It strikes me as an interestingly edible industrial material, ideally suited to be cut into strange shapes, artfully assembled and enthusiastically eaten. How would you go about machining toffee? Hot wire or waterjet, I'd have thought - I can't imagine a saw or a drill that would work nicely with swarf as gungy as toffee. You planarise the top with a flat teflon-coated surface and a weight while the toffee is still liquid.

I notice that someone appears to have managed to patent the mortice-and-tenon joint when made in gingerbread.

I have just emerged from a five-hour optimisation trance, fuelled by sushi and plum wine.

Here is the code; it looks for numbers which can be written as the sum of three sixth powers in two different ways. It's multi-threaded (by constants in the code assuming you have four cores spare) and uses a cache-friendly blocked linked-list structure; it slices up the problem into chunks by sum-modulo-P and then into buckets by sum-modulo-Q, then sorts the bucket contents. It uses moderately prodigious amounts of memory - about 20N^2 bytes. It takes 1m15s wall-time on my machine (2.4GHz quad-core) to run 'sumsix_t4 2003', 10m35s wall-time for 'sumsix_t4 4007'. I'm surprised that the output from 'time' indicates that it spends non-trivial time in the OS: how? The only bits that look like OS calls are print statements and memory allocation, and it only does O(N) of those.

real 10m35.586s
user 31m20.462s
sys 5m20.408s

I'm sure it could be significantly faster - it's trivially parallel and I'm only getting x3 speedup on four CPUs - but I am not quite sure how. I've looked at profiles, I suspect I may have been foiled by out-of-order execution, where quick-op ( result of slow-op ) has to wait for slow-op to complete and so appears as a hotspot when the real hotspot is elsewhere. Hard-wiring the values of 'N' and 'bucket' so that the compiler can replace the modulo operations with multiplies by magic constants doesn't make a difference.

If there are people out there who like this sort of challenge, I'd like some input.

Sage Ross, a Wikipedia editor, has carefully scanned in all hundred out-of-copyright plates of Haeckel's Kunstformen der Natur - a series of really beautiful turn-of-the-20th-century lithographs of attractively-shaped organisms from all the kingdoms of life, artfully arranged. With a pronounced bias towards the frondier kind of invertebrate.

Have some sea-slugs: