# Articles to 2019-11-10

First the link to this week's complete list as HTML and as PDF.

***

When charging a Li-Ion-cell at 6 C/h you wear it out in less than 20 full cycles. By heating the cell, charging it at 60°C, and discharging at room temperature Yang et al. managed to raise that one hundred fold to 2000 full cycles. Quite impressive, even if only done at the laboratory scale with single cells of 35 Ah or less. The heating itself also uses up an additional 5 % of the energy stored. Let's scale that up to the achievement claimed in their headline, charging for another 300 km of driving in ten minutes. For 300 km my new and efficient car uses 20 l of petrol or 177 kWh. Assuming 25 % efficiency for the combustion drive train and 100 % for electric we divide that four to get 44 kWh or a power of 265 kW for ten minutes. At an assumed 400 V that's 660 A of current. That current has to be supplied through a connection made by untrained personnel without using tools, i.e. a plug. Demanding that the heat generated at the connection must not exceed a quite generous 100 W, the contact resistance must not exceed 0.2 mΩ – and that after making and breaking 40 connections per day for months or years, not in a clean room but outside in the open. Do you see that happening out in the real world anytime soon?

***

Chan et al.'s complete model rests on the distribution of recent, i.e. current genomes alone. That does not, of course, make it wrong, but neither does it settle the question to the degree they try to imply.

***

Jacobs tries to explain quantitative methods to the humanities, which is of course a commendable endeavour. Unfortunately his exposition is riddled through with errors, beginning with his absurdly wrong figure 1. His main mistake, though, is the treatment of significance. What he does not seem to understand is that by definition one sample in 20 has to fall outside the 2 sigma range. Those are the ones he calls “abnormal”. In his examples he looks at many different linguistic features – not quite 20 in his short explanatory samples but easily that many in real life. Picking the one in twenty outliers and proving them to be abnormal at the five percent significance level is nothing more than tautological circular reasoning. Which is a shame, because when done correctly statistics can really help in testing an argument, mostly in the negative sense. Given the rare occurrence of some putatively diagnostic features in the text, a seemingly striking dichotomy may turn out to be well within what is to be expected by chance alone. The example he gives for just that is the best and most valuable part of his contribution.

***

Anthropogenic climate change is an undeniable fact and Maxwell et al. give another example of the real problems that can't be solved by windmills and wasting the last natural gas reserves.