Wikipedia:Reference desk/Archives/Science/2017 February 26

Welcome to the Wikipedia Science Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

February 26

Stereochemistry

This question is the result of reading Oxcarbazepine and Carbamazepine, and noticing from the infobox imagery that O... seems to be C... with an extra oxygen attached to the central ring.

How do we determine the specific structural formulas of not-so-simple molecules? I remember from high-school chemistry that the empirical formula is derived from breaking down the molecule (e.g. electrolysis of water yields twice as much hydrogen as oxygen), and I suppose that the basic chemical formula can be derived from a ratio of the empirical formula to mass (e.g. hydrogen peroxide will be twice as massive as peroxide, despite producing the same hydrogen/oxygen ratio), but neither one tells us anything about the shape. Based on the comments about Pasteur in the stereochemistry article, I'm guessing that the theoretical shape is the result of directly observed physical properties, but with molecules the size of these two anticonvulsants, I wonder how adding one oxygen would produce a molecule with distinctly different physical properties that can be determined to be independent of its chemical properties. The article refers to Stereoisomerism, so maybe my answer is there, but the article's technical enough that I don't understand it. Nyttend (talk) 12:30, 26 February 2017 (UTC)

You're correct that "total molecular weight" and "empirical formula" together can be used to solve for the chemical formula (fairly simple algebra if you have good-quality data). For the connectivity of the atoms and the conformation, IR, NMR and X-ray crystallography are the three classical tools one uses. DMacks (talk) 15:18, 26 February 2017 (UTC)

Also Mass spectrometry. --Jayron32 22:45, 26 February 2017 (UTC)

Just molecular weight and empirical formula usually can't solve the structure, see isomers. Fgf10 (talk) 08:14, 27 February 2017 (UTC)

@Nyttend:, so far we are answering about the chemical structure (approximate 3D location of each atom and to which other one(s) it is bonded). Are you also asking about actual physical properties such as density, melting point, optical rotation, color, etc? Or biochemical properties, such as ADMET, receptor target specificity and affinity, actual cellular or systemic effects, etc? DMacks (talk) 23:00, 26 February 2017 (UTC)

I got the impression that the beginnings of stereochemistry came from Pasteur's observation that different version of the same compound had different physical properties, which I'm guessing means that he produced some theory (still followed in its basic idea) that the two had identical chemical formulas but different shapes. That's the only reason why I mentioned the physical properties. I'm trying to understand how we get a sense of the structure; I don't have time to check the links that you and Jayron provided above (in a fast-food restaurant at the moment), but I'll look at them when I get home. Nyttend (talk) 01:27, 27 February 2017 (UTC)

My recollection from undergraduate organic chemistry is that in the days that these things were invented, the complicated structures were derived from "proof by synthesis". That is, if you can synthesize the molecule using well studied reactions that yield predictable products, you can show that your molecule must have a particular structure, given the synthesis protocol. You can further confirm this by decomposing the structure, again using predictable reactions, and showing that the expected products result. I haven't been able to find any particular reference work describing how to do this, though you'll find lots of papers from the 50s and 60s doing a particular molecule by searching for "proof by synthesis" and similar terms. Nowadays I guess you just stick the thing in an NMR and call it a day. Someguy1221 (talk) 08:46, 27 February 2017 (UTC)

Absolute configuration was not known until 1951. Apart from this, much of the rest was clever interpretation of the recognizable effects of various bond stretches (for infrared spectroscopy), magnetic coupling between nuclei (nuclear magnetic resonance), and blasting apart molecules in smaller chunks (mass spectroscopy). Once X-ray crystallography was known, it became a great fall-back tool for whatever leftover dammit-I-just-can't-figure-this-out spectra researchers might run across. Wnt (talk) 17:13, 2 March 2017 (UTC)

How to hasten the separation of oil from peanut butter?

I made my own peanut butter by grinding up nuts but it's too thin. Apparently if you let it sit for a long time, the oil rises to the top and can be poured off, leaving something more viscous. How could this be easily hastened? Would vibration on top of a washing machine make it better or worse? I can't really strap it down inside the washing machine. ----Seans Potato Business 12:52, 26 February 2017 (UTC)

Anyone who would even think about taping a jar of peanut butter to the drum of a washing machine is a kindred soul. More practically, you might use a long piece of hosiery (or similarly shaped apparatus) as a makeshift centrifuge. Put the jar inside it oriented so that the bottom of the jar is at the bottom end of the sock or whatever, then swing it round and round. Shock Brigade Harvester Boris (talk) 15:20, 26 February 2017 (UTC)

I googled for [how to make peanut oil] and got lots of hits for several different methods at varying levels of sophistication or equipment. DMacks (talk) 15:28, 26 February 2017 (UTC)

Stick it in the refrigerator for a couple of days - it will thicken up. You have squeezed oil out of the nuts when grinding them - but the solids will reabsorb much of the oil. Wymspen (talk) 15:49, 26 February 2017 (UTC)

You could keep it in a sieve or a suspended muslin bag (jam/jelly bags are ideal) - the liquid oil will drip out. You probably want to do this at around 10°C - too cool and the peanut oil will wax; too warm won’t prevent it going rancid. LongHairedFop (talk) 16:50, 26 February 2017 (UTC)

• Try squeezing the bag to hasten it further (have paper towels ready, but you will need to wash hands with soap to get them clean).

• Also, the oil does become thicker at lower temps, so, if eating it cold is an option, then that approach will work.

• Finally, if you intend too use the peanut butter in a recipe with butter, margarine, or oil, like baked goods, you could just reduce the amount of those. StuRat (talk) 17:34, 26 February 2017 (UTC)

A Centrifuge is the most common technical approach for such separations. This was already pointed out in the first post. --Kharon (talk) 11:04, 27 February 2017 (UTC)

I was thinking maybe that you could hit the peanut butter with ultrasound and "shake out" the solids. (I was thinking that some kind of 'liquefaction' similar to the New Madrid earthquake might be accomplished on microscopic scale) From some searching it appears that ultrasound-assisted extraction of some plant oils is indeed a thing, but it is used to persuade oil, protein, or other components to enter a solvent; the solvent is then further processed. [1][2] My initial thought to more or less scale down the sound wavelength with the particle size was probably stupid to start, but I might as well mention what I saw, though I strongly doubt home producers want to play with hexane or even ethanol evaporation just to get peanut oil. Wnt (talk) 09:07, 1 March 2017 (UTC)

How far ahead can the weather be predicted with reasonable accuracy?

With the best models in use today for weather forecasting, how far ahead can the weather be predicted with reasonable accuracy? --100.34.204.4 (talk) 14:15, 26 February 2017 (UTC)

Weather forecasting introduces the subject and Atmospheric model teamed with Numerical weather prediction details current techniques. Blooteuth (talk) 14:28, 26 February 2017 (UTC)

It depends what you mean by "resonable accuracy" as well as what part of the forecast and for what geographical areas, but the answer most people give is something from 48 hours (or perhaps next day) to 5 days although some say up to 9 or 10 days [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]. There seems to be general agreement anything beyond 10 days is mostly useless. Nil Einne (talk) 14:42, 26 February 2017 (UTC)

It also depends on the current weather situation. Its a lot easier to predict weather if there is a stable situation than in a case of a quick sequence of alternating high and low pressure systems coming in. --Stephan Schulz (talk) 14:39, 26 February 2017 (UTC)

I made some changes after you replied, adding references and extending the period to up to 9 or 10 days and mentioning more that more than 10 days isn't generally considered useful. [14] Hope you don't mind as this doesn't I think affect anything you said. Nil Einne (talk) 14:44, 26 February 2017 (UTC)

No worries. --Stephan Schulz (talk) 15:00, 26 February 2017 (UTC)

Also the Huffingtonpost reference suggests predictions during winter are less accurate (I didn't look at it carefully so I'm not sure if this is for reasons like the inaccuracy of snow predictions or other such complications). Nil Einne (talk) 14:53, 26 February 2017 (UTC)

Nil, you're a smart guy and I agree with you 99.9% of the time, but -- Mother Mary an Jozef, you're using teh Huffington Post as a source for a scientific topic? Here are verification stats from ECMWF. You can drill down for lots of details, including comparisons to other centers. Here are stats from NCEP. Notice the numerical forecast centers tend to focus on things like 500 mb anomaly correlation, which are well suited to basic questions of predictability and prediction skill but aren't of much interest to the general public. Here NCEP has some statistics on surface weather elements. Shock Brigade Harvester Boris (talk) 15:09, 26 February 2017 (UTC)

That NCEP stats page is pretty awesome, thanks! But I don't fully understand it. Can you tell us how we can use it to confirm, deny, or even just investigate the claim that winter predictions are in some sense less accurate than those in other seasons? SemanticMantis (talk) 16:12, 26 February 2017 (UTC)

First you need to decide what "less accurate" means, which can be a non-trivial problem in itself. ECMWF stats for 500 mb anomaly correlation outside the tropics shows that winter predictions are more accurate than summer predictions. It's pretty well accepted that summertime forecasting is harder than for winter, because summertime weather is dominated by small-scale phenomena such as convective storms rather than larger, long-lived frontal systems and the like. (That's why I went marginally sub-orbital about the HuffPo article.) Shock Brigade Harvester Boris (talk) 16:34, 26 February 2017 (UTC)

Apologies. I did include one research article and and one paper from a NOAA workshop but put these at the end because I get the feeling they are probably beyond the OP's interest. (Perhaps I'm wrong, but felt the limited info in the question suggested they weren't going to look into this in great detail, especially in to the technical science.)

Most of my other refs are targeted at a more general audience. Several (including the Huffington Post) seem to be small scale data analyses by people who I think at least have a decent understanding of statistics although they suffer from problems like generally very small sample sizes (generally 1 year or less and in one location) and the issues of handling different measures. Most of them seem to be using consumer data sites instead of government ones. One of them does note that these are generally less accurate something which I believe is well supported, when I was answering the question I forgot that the OP has specifically asked about the best forecasts so was thinking they would be interested in such sites. Other refs are quoting meteorologists or research (which is very risky given how wrong the media can get things) or blog posts from meteorologists (which doesn't necessarily mean much although it was on AGU). I did try to verify that winter claim, and came across [15] which seems to be a student project, broken perhaps because of age, which makes the same claim. I decided it was too low quality to include but mistakenly did let it influence my decision to repeat the claim.

I'd note looking more closely the Huffington Post article only seems to be looking at precipitation. Whether even this is generally true, I don't know.

Nil Einne (talk) 08:33, 1 March 2017 (UTC)

Lots of detailed statistics from the UK Met Office here - http://www.metoffice.gov.uk/about-us/who/accuracy/forecasts - making the interesting point that the four day forecast now is as accurate as the 24 hour forecast was 30 years ago. Wymspen (talk) 15:53, 26 February 2017 (UTC)

[16] which is linked in the Arstechica article above says "forecast skill in the range from 3 to 10 days ahead has been increasing by about one day per decade: today’s 6-day forecast is as accurate as the 5-day forecast ten years ago, as shown in Fig. 1" Nil Einne (talk) 08:55, 1 March 2017 (UTC)

I predict that on the 10th of November it won't rain here. Count Iblis (talk) 08:48, 1 March 2017 (UTC)