Buoy 300234062785480 air temperature: http://gergs.net/wp-content/uploads/2015/12/Buoy_300234062785480.png Temperatures were close to zero back in November, but that's not midwinter!

The volumes that Wipneus obtained from the CS-2 data team are plotted here: http://gergs.net/wp-content/uploads/2015/04/PIOMAS1.png Pretty close, except mid-winter. Details here: http://gergs.net/2013/07/northern-sea-ice-update/piomas2/

Volume: tks Wipneus; method?

"This result could not be produced by random factors." Not quite. The sample size is just one election. You're postulating that, for the first few groups on the ballot, those nearer to the left edge get a progressively higher vote on iVote than in the general count. Effectively you're suggesting that, for the first few positions, there's a trend in the difference between iVote and the general vote depending on the group position. That's indeed how it looks: http://gergs.net/wp-content/uploads/2015/04/iVote.png If we said that the trend extends out to 6 groups so that it gets down to zero somewhere between groups E and F, then it actually looks to be statistically significant (p ~ 2%). But -- and it's a big but -- that would be serious cherry picking; there'd be lots of other "single sample" anomalies that might have drawn our attention here, but aren't considered in such a significance estimate. Even if the trend were just as postulated (6 groups long including the zero at the end), the probability of seeing a trend that big by chance alone is about 1 in 50, if the underlying process ("i-voting") contained no trend difference at all. There'd be better ways of estimating the probability of seeing something like what you've highlighted, but they would need additional data. Maybe you know someone who could have a proper go at it ... or maybe we just wait for a few more "samples". COMMENT: The nice trend line on your graph is invalid as you miss two-thirds of the ballot paper columns. I also included the Labor, Liberal/National and Green data as we have long experience of knowing these parties have differences by vote type, yet you've included these parties in the trend line and I don't think they should be there. You are better to use groups where there is little variation by vote type. It is fascinating to compare groups D and H, groups with minimal support. D has an iVote surge but H doesn't. When I tried iVote there were 6 columns displayed on the screen, A to F, which I think help explains the finding that the effect drops after 6 columns. Maybe there is a 1 in 50 chance of this being produced randomly, but no one will take the risk again. There's enough in this one example to suggest electronic ballot papers should be presented to voters in a different manner. If you really wanted to test the statistical probability. the solution is to use more data points. Compare iVote by electorate and see if the pattern is consistent. Compare the iVote support for a low polling party with its support by poling place across the state.

Nnnnevin ... more n's than any other network, or at least more northern nous: https://www.youtube.com/watch?v=Q9GZuT3HN6Y

Plotted above of course (with nice star field), but it's perhaps worth remembering what late summer Antarctic sea ice actually looks like (from ClimateReanalyzer.org): http://pamola.um.maine.edu/DailySummary/frames/GFS-025deg/DailySummary/GFS-025deg_NH-SAT5_SEAICE-SNOW.png That is that there's hardly any ice at all except in the lee of the Peninsular. The source of the variability is pretty obvious, and the likely source of the increase in variability is not hard to glean.

The August 1922 DNI sea ice chart is here: http://brunnur.vedur.is/pub/trausti/Iskort/Jpg/1922/1922_08.jpg ...and in context here: http://gergs.net/2013/07/more-northern-sea-ice/1920-1939/ The red sea ice markings north of Spitsbergen might even be Hoel's assuming the Norwegians and Danes shared data. The marks indicate clear water extending to about 81.5°N, as do other markings immediately to the east. There was no satellite, and air survey is not likely, so these records are probably from close ship survey.

Thanks; here's yet another (very slightly different) view: http://gergs.net/2013/07/northern-sea-ice-update/