4 ms·

probably a minor and tedious point, but it bugs me when journalists do this: > (3) The universal force of gravity is very stable. Newton's gravitational consta

by brey 13y ago

probably a minor and tedious point, but it bugs me when journalists do this: > (3) The universal force of gravity is very stable. Newton's gravitational constant G has changed less than 1 part in 100-billion since the laser experiments began. technically worded correctly, but this language implies that we know it IS changing, by some small but non-zero amount. actually, we haven't detected any change at all - so there's every chance it's NOT changing - but if it is, it's changed less than this small amount which our instruments won't measure more accurately than. > No variation of the gravitational constant is discernible, (dG/dt) / G = (0.0±1.1)x10–12 /yr http://ilrs.gsfc.nasa.gov/docs/williams_lw13.pdf http://ilrs.gsfc.nasa.gov/docs/williams_lw13.pdf

hartror 13y ago
Came to the comments looking for this exact explanation as the implication in the article struck me as strange. Thanks! :)
- brey 13y ago
  If we had detected some non-zero change in G, that would be big news.
randomvisitor 13y ago
> No variation of the gravitational constant is discernible, (dG/dt) / G = (0.0±1.1)x10–12 /yr If we're obsessing over details, I'm wondering if 0.0 is indeed the center of the confidence interval, or if this is just a way to say they have bounded |G'/G|. The cited paper ("Lunar laser tests of gravitational physics") just states the result without giving much detail.
- ISL 13y ago
  In general, because G' is a signed quantity, I would expect any experiment to be quoting double-sided errorbars. If G increases, one thing happens, if G decreases, the opposite happens, so the difference is discernable. I'd take 0.0 to be their reported central value.
joshuahedlund 13y ago
Can you explain how that reconcile with this recent news?[0] "Their tests yielded a new G value of 6.67545 × 10−11 m3⁄kg s2, which is higher than the current accepted value by about 240 parts per million" Of course, that particular experiment could just have had some corrupting factors, but how do you know what experiments support a constant value and which ones don't without approaching tautology? [0] http://www.wired.com/wiredscience/2013/09/high-gravitational-constant/ http://www.wired.com/wiredscience/2013/09/high-gravitational...
- VLM 13y ago
  Short version is you're talking about a value, they're talking about a first derivative of that value. I can take two COTS cheap canned 64 MHz oscillator modules like the ones on your computer motherboard, and the mfgr claims they'll output something within 100 ppm of 64 MHz. Once the temp stabilizes they'll individually be WAY more stable than 100 ppm comparing right now to an hour ago, but whatever it is they oscillate at, it'll be within 100 ppm of each other. If you graph the output of NTP, perhaps via Munin or any number of other sysadmin tools, you'll see most of those devices are very stable from hour to hour, at a basically fixed offset from the correct frequency. A couple years ago there was a ham radio construction project in QEX etc along the lines of using a crude computer grade COTS osc, feeding it thru the equiv of a hundred-something times multiplier, and using it as a local oscillator. I built a couple and they're ridiculously stable, although all over the map. I don't have an 1152.000000 MHz oscillator although from memory they were all within 10 or so ppm of each other (better than the rated 100ppm) Edited to add, if you're running Debian Linux and NTP, take a look at /var/lib/ntp/ntp.drift. The generic COTS dell 6-U server one of my images runs on, ticks very stable and precisely -40.069 ppm low compared to reality and it never varies. On the other hand, my "desktop" running basically the same software but on generic dell desktop hardware is at present running +9.615 ppm fast at this instant although it varies a bit more (probably due to temp swings).
- dalke 13y ago
  G is notoriously hard to measure accurately. Wikipedia's entry on the topic suggests that you read a 1997 review paper at http://iopscience.iop.org/0034-4885/60/2/001 http://iopscience.iop.org/0034-4885/60/2/001 . That abstract starts: > Improvements in our knowledge of the absolute value of the Newtonian gravitational constant, G, have come very slowly over the years. Most other constants of nature are known (and some even predictable) to parts per billion, or parts per million at worst. However, G stands mysteriously alone, its history being that of a quantity which is extremely difficult to measure and which remains virtually isolated from the theoretical structure of the rest of physics. Several attempts aimed at changing this situation are now underway, but the most recent experimental results have once again produced conflicting values of G and, in spite of some progress and much interest, there remains to date no universally accepted way of predicting its absolute value. The Wired article confirms that it's hard to measure, with some examples of the difficulties. However, G* M is not hard to measure accurately. We believe the Earth gains about 100,000 kg per year from meteors, and in any case, the Moon and the Earth have changed by much less than one part in a million over the last few decades. So any change in G* M should be due to a change in G, even if we can't measure G directly. This is of course also why we don't know the mass of the Earth better than a few hundred parts per million.

by brey 13y ago

hartror 13y ago

Came to the comments looking for this exact explanation as the implication in the article struck me as strange. Thanks! :)

brey 13y ago
If we had detected some non-zero change in G, that would be big news.

randomvisitor 13y ago

> No variation of the gravitational constant is discernible, (dG/dt) / G = (0.0±1.1)x10–12 /yr If we're obsessing over details, I'm wondering if 0.0 is indeed the center of the confidence interval, or if this is just a way to say they have bounded |G'/G|. The cited paper ("Lunar laser tests of gravitational physics") just states the result without giving much detail.

ISL 13y ago
In general, because G' is a signed quantity, I would expect any experiment to be quoting double-sided errorbars. If G increases, one thing happens, if G decreases, the opposite happens, so the difference is discernable. I'd take 0.0 to be their reported central value.

joshuahedlund 13y ago

Can you explain how that reconcile with this recent news?[0] "Their tests yielded a new G value of 6.67545 × 10−11 m3⁄kg s2, which is higher than the current accepted value by about 240 parts per million" Of course, that particular experiment could just have had some corrupting factors, but how do you know what experiments support a constant value and which ones don't without approaching tautology? [0] http://www.wired.com/wiredscience/2013/09/high-gravitational-constant/ http://www.wired.com/wiredscience/2013/09/high-gravitational...

VLM 13y ago
Short version is you're talking about a value, they're talking about a first derivative of that value. I can take two COTS cheap canned 64 MHz oscillator modules like the ones on your computer motherboard, and the mfgr claims they'll output something within 100 ppm of 64 MHz. Once the temp stabilizes they'll individually be WAY more stable than 100 ppm comparing right now to an hour ago, but whatever it is they oscillate at, it'll be within 100 ppm of each other. If you graph the output of NTP, perhaps via Munin or any number of other sysadmin tools, you'll see most of those devices are very stable from hour to hour, at a basically fixed offset from the correct frequency. A couple years ago there was a ham radio construction project in QEX etc along the lines of using a crude computer grade COTS osc, feeding it thru the equiv of a hundred-something times multiplier, and using it as a local oscillator. I built a couple and they're ridiculously stable, although all over the map. I don't have an 1152.000000 MHz oscillator although from memory they were all within 10 or so ppm of each other (better than the rated 100ppm) Edited to add, if you're running Debian Linux and NTP, take a look at /var/lib/ntp/ntp.drift. The generic COTS dell 6-U server one of my images runs on, ticks very stable and precisely -40.069 ppm low compared to reality and it never varies. On the other hand, my "desktop" running basically the same software but on generic dell desktop hardware is at present running +9.615 ppm fast at this instant although it varies a bit more (probably due to temp swings).
dalke 13y ago
G is notoriously hard to measure accurately. Wikipedia's entry on the topic suggests that you read a 1997 review paper at http://iopscience.iop.org/0034-4885/60/2/001 http://iopscience.iop.org/0034-4885/60/2/001 . That abstract starts: > Improvements in our knowledge of the absolute value of the Newtonian gravitational constant, G, have come very slowly over the years. Most other constants of nature are known (and some even predictable) to parts per billion, or parts per million at worst. However, G stands mysteriously alone, its history being that of a quantity which is extremely difficult to measure and which remains virtually isolated from the theoretical structure of the rest of physics. Several attempts aimed at changing this situation are now underway, but the most recent experimental results have once again produced conflicting values of G and, in spite of some progress and much interest, there remains to date no universally accepted way of predicting its absolute value. The Wired article confirms that it's hard to measure, with some examples of the difficulties. However, G* M is not hard to measure accurately. We believe the Earth gains about 100,000 kg per year from meteors, and in any case, the Moon and the Earth have changed by much less than one part in a million over the last few decades. So any change in G* M should be due to a change in G, even if we can't measure G directly. This is of course also why we don't know the mass of the Earth better than a few hundred parts per million.