Is it possible to detect disparities of the human eye, depending on the observers? If true, what would be an unpersonal device, in order to analyse the spectral answer of the human eye?
(This set of ideas results from reflections between two AAVSO observers, DMIB and PREB, after a common observation night)
1 _ What happens?
During a short amount of time (15mn), the visual measures on the same star show, in the AAVSO database, a spread of 0.4 to 0.8 magnitude. Much more as what one expects from the 2 objective factors:
_ ability of an observer to estimate a 0.1 magnitude spread.
_ reference stars on AAVSO maps are given with an accuracy of 0.1 magnitude.
When an observer locates a variable star 'in the middle' of 2 references 7.24 and 7.44, he will read '72' and '74' on the AAVSO map, and then he will publish '7.3'; when the same observer locates the same star 'in the middle' of 2 other references 7.26 and 7.46, he will read '73' and '75' on the AAVSO map, and then he will publish '7.4'. Giving a 0.1 magnitude spread.
_ measures worsening cases are sufficiently rarely noticed, and then neglected (light pollution, moon, light clouds, plane tracks, low in the sky, near observing limit, observer's tiredness).
2 _ Measure of the spectral signature of the eye's sensitivity.
Design a device, which, for a given wavelength, simulates a stars field on a black background, with an extra light source, shaped as a little ring and which provides a sequence of increasing magnitudes, by 0.1 magnitude steps. Using this device, the observer notices at which step of the sequence, he detects the ring (its place is unknown and random).
A similar sequence is provided for a scale of several wavelengthes.
One gets a curve, giving sensitivity threshold depending on the wavelength: the observer's signature.
3 _ Results and questions.
Giving a large sample of variable stars visual observers, which is the most likely spectral signature? Is there an important spread for some wavelengthes (red for instance)?
The sensitivity threshold is not the same, with the use of different instruments (naked eye, binoculars, Schmidt-Cassegrain telescop with or without star diagonal). Is there a change in the spectral signature, depending on those optical add-ons?
4 _ Other influences.
The Purkinje effect (red stars) needs to use the device only with out-of-axis vision.
Some magnitude measures are done, looking the sensitivity threshold by gradual unfocusing of the instrument. Is there the spectral signature affected by such a practice?
5 _ Design of the physical device.
(It cannot be a software on a laptop, due to the poor scale of magnitudes available on a screen, and its non-logarithmic nature).
Todo...
Are you candidate? Any suggestion?
Pierre
Pierre, let me offer just a couple of comments concerning your post addressing observer accuracy.
While I, too, regard an observational scatter of 0.4-0.8 magnitudes between assumed experienced observers to be disturbingly large, your results fly in the face of those from my own tests conducted years ago at various gatherings of AAVSO members. One such test that particularly stands out in my memory was one that involved several of the AAVSO's leading observers at the time (Peltier, Cragg, Lowder, plus one or two others and myself). The group employed the same telescope at the same location on the same night. After conducting a series of estimates of the brightness of variables in several fields we all compared notes. What we found was that we did not deviate from one another by more than 0.1 magnitudes. I conducted other such experiments on later occasions always with essentially the same results.
I would also caution placing too much confidence in the absolute visual accuracy of CCD-V magnitudes, especially to two decimals, when employed to define visual scatter between observers. CCD-V, even with today's correction factors applied, often does not correspond precisely to what the observer's eye may see. Stars whose spectra showing the least abnormalities from the norm may give skewed CCD-V magnitudes relative to true visual. To be very honest I rarely see AAVSO charts where every star in a CCD-V sequence appears to my eye to fit perfectly. There often seem to be one or two that don't quite fit and in some instances the deviations seems rather dramatic, amounting to several tenths of a magnitude.
I certainly encourage you conducting further experiments, but would suggest involving a larger group of individuals than just two as there certainly are small consistent differences between some observers. I would also suggest choosing a selections of variable star fields with CCD-V magnitudes that the group beforehand agree seem to be visually consistent in the accuracy of the listed brightness of the comp stars.
J.Bortle (BRJ)
Hello, I am in agreement with John Bortle, that good visual estimates by careful observers made simultaneously shouldn't differ by more than 0.1 magnitude, assuming the same comparison stars are used. If you are experiencing differences of 0.4-0.8 magnitudes using the same comparisons, at the same moment of observing, something is seriously wrong!
Some possible explanations:
1. Clouds. If there are some clouds moving around, it can cause significant variations due to the density of the clouds over short distances can make the variable and comparison appear much differently.
2. The variable and/or comparison have large color differences or emission lines. Such large variations can affect different observers in a significant way. Best way is to use good constant, "normal star" comparisons of B-V <1.0 Some "pathological" variables of extreme color or type (like V838 Mon) are very difficult to achieve consistent results between observers, and CCD too.
3. The telescope optics have some issues. Sometimes dirt or fog on the eyepiece or secondary can affect one part of the FOV more than the other. Also you could inadvertently hit the "blind spot" in the eye. Its important to look at the variable and comparison from a couple of different angles or positions, and "average" your results.
4. Try to keep variable and comparison the same distance from optical axis. Field vignetting, or off-axis aberrations can affect the brightness, if the two things are in different parts of the FOV. (Eg. in a Newtonian, two stars of the same brightness would appear different if one was in the center , and one was near the edge due to coma spreading and diffusing one star)
5. Defocussing is very helpful if the star is bright enough. Its easier to compare disks brightness than very tight points. It also lessens the effect of aberrations and scintillation.
6. Make sure no misidentification of variable or comparisons! (eg. if close neighbors, insufficient resolving power, combined flux, etc.)
In general, the difference between people's spectral response is very small. There are people who have significant genetic faults in their visual receptors (eg. partial color blindness), but those should be known and obvious. The vast majority of "normal" observers spectral response curves will be very much alike, so attempting any corrections for such miniscule variations would likely be a waste of effort.
Mike
Thanks for your answers! What appeared me before as confused and unclear becomes close to convictions!
Yes, I agree with John: same location, same instrument, same time (+/-30mn), same comparison stars, that drives to a deviation of 0.1 magnitude. Provided that the spectrum of the variable star is not too abnormal. I keep in mind from John's experience, that for a normal star, most of the people show a common response of their eyes. But a doubt remains: with a (very) large sample of observers, is it possible that a little fraction of this sample shows a systematic deviation?
In the same manner as the daltonism which affects very few people.
I agree too, there is no chance that the spectral responses of a CCD and of the human eye may always fit.
Thanks to Bob's informations about the asymmetry of the retina. I ignored that before. But I am always careful with my own eyes, like with any physical device: when measuring out of focus, comparison stars and variable stars are placed in the same angular sector with regard to the look direction. And if possible with binoculars, I repeat the measure with 2 attitudes (lying and sitting), so that the stellar field rotates. That answers also the Mike's topic #3.
And now, thanks to Mike.
_ Yes, DMIB and me are very careful about light clouds. Because of that, I repeat a measure throughout half of an hour. To be sure, but who knows?
_ Your topic #2 has been the ground idea of my initial reflection. Novae are often reddening after some time...
_ No problem with the topics #4, #5, #6. Just about the #5, I suggested that the spectral response may differ, because of the cone and rod cells distribution.
_ About the last topic, the question is: how to depict the no-man's land between "normal" observers and "pathological" observers? In terms of severity and frequency.
After all, we speek about the receiver. But what about the emitter itself?
I remember, end of august 2013, a look towards the light curves of NovaDel2013 gave me the idea of writing a program in order to detect quick periodic variations. I wrote it, even I published it
("http://etoilesvariables.fr/index.php?post/2013/11/15/
NovaDel2013-a-t-elle-des-variations-de-l-ordre-de-l-heure"), but never found anything. For a given wavelength, too few observers. And worse, a big daily void of observers over the Pacific Ocean...
Strangely, at the same time, Arne Henden opened a discussion on possible "hourly variations"!
This year, I followed NovaSgrNo.2 and AGPeg. And always the same question...
Pierre
The sharp vision (daylight) cone cells are heavily concentrated in the central .5 deg of the field of vieiw. Night vision rod cells are more concentrated in an oval 10-20 deg away, but there is a creater density of these in a crescent shape in the upper part of the oval (this means below the center of the field of view). This could effect you if you're using a large out of focus image, but I think also with stars in focus. Bob Young