Variable water flow is toward the laser end.
Something I have perhaps not made very clear in the past is that the experiment is in no way designed to confirm Fresnel's prediction or the conclusions drawn from Fizeau's ether drag experiment.
The purpose of the experiment was to detect a fringe shift, not to precisely describe how much it would be. Since nothing is changing while the device is turned between east and west, the only possible errors will stem from friction in the water connection, and leveling of the apparatus. But since both paths are always equally affected, the result would still be sound.
The setup used for the experiment was similar to the Fizeau's ether drag experiment, but with major variations in that there's only a single straight tube carrying the liquid flow and the apparatus can be rotated to point east or west.
The setup was designed thus.
Variable water flow is toward the laser end.
The light path from the laser is split at the first mirror, which is semi silvered. One beam travels through to the end of the water filled tube where it's reflected via two mirrors to return along the air path beside the tube. It's then reflected at 90 degrees to again pass through the semi silvered mirror and finally arrive at the viewing screen, after being magnified through a lens. The other half beam travels the same path but in the opposite direction, and is reunited with its counterpart to follow along the same path direction after it's finally reflected off the semi silvered mirror.
The apparatus is extremely stable because the light paths are almost identical. Temperature variations are of no consequence for that reason.
Turning the device while the water flow was stationary had no effect on the interference pattern. The fringes only shifted when the water was flowing, and the shift was proportional to that flow rate, as was the case in the Fizeau experiment.
Changes to the water flow rate must distort the light paths to some degree and could show up in the interference pattern, but only when the flow rate is changing. But the flow rate remains constant while the device is turned.
Assuming that the frame of the device is the base on which light
propagates.
With the laser source on the west end of the apparatus and the water
flowing west toward the laser, as in the above diagram, the propagation
base is dragged along with the water according to wv*(1-(1/n^2))
(wv = water velocity). An interference pattern is generated on the screen
via the split light paths.
Turning the device 180 degrees so that the laser source is now on the east end and the water flow is pointing east, toward the laser, the propagation base is dragged along with the water according to wv*(1-(1/n^2)). Exactly the same interference pattern is generated on the screen. Nothing has changed. If there is a pattern change, it's caused by some mechanical means as the device is rotated, and the cause would be very obvious.
Assuming now that the ECI frame is the propagation base for light.
With the device pointing in the north-south plane, any chosen fringe
will be the focal point around which it will shift when the water flow
is pointed east or west. The position of the focal point shifts across
the screen as the water flow rate is varied, but it is still the focal
point for the fringe shift when the flow rate is constant and the device
is rotated. Increasing the water velocity causes the interference pattern
to shift in a particular direction and, with the water flow now constant,
the pattern changes in that same direction when the device is turned
toward the east, as is expected.
Example: Assuming that light propagates according to the non
rotating frame of the earth (the ECI frame).
c = 3E+08
v = 400 Tangential velocity for Melbourne Australia
wv = 6.1 Water velocity per second (= .1 wavelength shift)
n = 1.332 Water/air refractive index.
l = 3.6 The water tube length is 1.8 meters. But both beam
paths are affected.
(for a water velocity of 6.1 m/sec)
The orientation of the device is set so that the water flow is
pointing west. Relative to the ECI frame, the water is traveling
at v - wv = 393.9 m/sec. The propagation base on which light
travels is dragged (v - wv) * (1 - (1/n^2)) = 171.8875 m/sec away
from the ECI frame.
The orientation of the device is now set so that the water flow is
pointing east. Relative to the ECI frame, the water is traveling
at v + wv = 406.1 m/sec. The propagation base on which light
travels is dragged (v + wv) * (1 - (1/n^2)) = 177.2112 m/sec away
from the ECI frame.
The difference is; 177.2112 - 171.8875 = 5.323754 m/sec of water
path length: 5.323754 / c * 3.6 = 6.388505E-08 meters for a 3.6
meter water path length, which is .1007651 of a wavelength.
If the water flow direction is reversed relative to the laser source,
the focal point for the fringe shift will move to another spot in the
opposite direction across the screen. There will be no problems if I
choose to run the test with this new configuration, so long as I don't
change anything while the device is being rotated.
Ignore the surrounding clutter in these images. Removing the background and gold plating the apparatus won't have any effect on the result.
Prior to fitting the tube, its windows were roughly aligned square to the tube by reflecting a laser pointer off the window face while the tube was being rotated along its axis. Any misalignment caused the reflected spot to scribe a circle. Each tube end was adjusted until the reflected spot remained still. The final adjustment was done by slightly bending the (now water filled and in place) tube with the adjusting screws until each half beam found its way to the center of the far end of the tube.
If the windows are not aligned perfectly square to the beam paths the paths will, due to refraction, never be the same, and a stable interference pattern would be impossible.
Before the water tube was fitted, all laser paths were aligned. The mirror setup naturally caused feedback directly into the laser output mirror, and that sets a very clear and very rigid interference pattern traveling in both directions along the beam paths, right up to the screen. It would be expected that the phase relationship between the laser and the feedback beam would be locked together. But when the water filled tube is in place, the intensity of the beam traveling through the water is substantially reduced. The greatest feedback is then generated from the windows at each end of the tube. The phase locking can't involve every component in the system at the same time.
Two sets of interference patterns appear on the screen. One is caused by feedback and is completely rigid. To avoid confusion, the fringes of the two sets can be arranged so that they are at an angle to each other.
The torsion force acting on the swivel, applied in either direction, was easily shown to be of no consequence to the interference pattern.
Vibrations from the twin screw pump are absorbed in an air bubble housed in a football bladder.
The maximum pump pressure used was 200 psi, when the pump was running at around 5000 rpm. 100% efficiency would give a water flow rate of 51 meters per second. Judging by the fringe shift per Freznel's prediction, it was about 20 m/sec.
This detail on how the experiment was conducted should demonstrate that the evidence for light speed anisotropy was not just a first impression. My method may seem a little unorthodox, but the conclusion is sound.
Because space is continually recycled in my garage, the device was pulled apart and reassembled many times. The interferometer was not stored in a secure environment either. The mirrors endured several encounters with my cat's tail. But that was part of the test as well because for each time the device was tested, it was necessary to start all over again, removing the water tube, resetting the mirror paths, realigning the tube windows, fine tuning the windows after the water was connected and the unit attached to the turntable, which was also newly installed and leveled. The idea was to shift things about in an attempt to identify any flaws in the setup. But the fringe shift never failed to tell the same story.
Although, at setup number three, after determining which way the interference pattern would shift as the water flow rate is increased, to my disappointment, when the device was "pointed east" the fringes shifted in the opposite direction to what should have been. It was pointless using observer bias to try and conceal what was clearly shown. I just threw in the towel, realizing that I was apparently not seeing proof of a light speed anisotropy at all, only some fatal flaw in the apparatus that I couldn't possibly correct, or even identify, and that deemed the device beyond redemption.
It took me several hours to realize that when I "pointed the device east", I had pointed the beam from the laser source east, and not the water flow direction. That was pointing west. So the result was exactly according to plan.
The following is only included to demonstrate why no fringe shift occurs when the device is turned while the water is not flowing.
With the water flow stopped:
The beam is initially split at
the first mirror and one half travels the air path while the other half
travels the water path, to the far end. The unidirectional paths are all
that need to be considered here.
EXAMPLE (for any pointing direction).
WATER PATH c = 3E+8 wl = 634 Wavelength in nanometers. n = 1.332 Water refractive index. v = 400 Tangential velocity. pl = 1800000000 Path length in nanometers. Linear speed of light = 1/n = .7507507507507507 * c Water propagation center is dragged v*(1-(1/n^2)) = 174.5493240988737 m/sec east of the ECI frame. Beam source is 400 - 174.5493240988737 = 225.4506759011263 m/sec further east than the propagation center, relative to the speed of light with no medium. Relative to the speed of light in water, the propagation base is 225.4506759011263 / .7507507507507507 = 300.3003003003003 m/sec to the west. The distance the beam must travel along the water path increases pointing east and decreases pointing west, at the rate of 300.3003003003003 meters per second of path length. Number of wavelengths along the path if the device is fixed with the ECI frame = pl/(wl*1/n) = 3781703.470031546 Number of waves along the path with the device moving east at 400 m/sec and the beam pointing east = (((c+((v-(v*(1-(1/n^2))))/(1/n)))/c)*pl)/(wl*1/n) = 3781707.255520505 which is 3.785488958936185 more waves than the ECI frame fixed device. Number of waves along the path with the beam pointing west = (((c-((v-(v*(1-(1/n^2))))/(1/n)))/c)*pl)/(wl*1/n) = 3781699.684542587 which is 3.785488958936185 less waves than the ECI frame fixed device. AIR PATH n = 1.0003 Air refractive index. Linear speed of light = 1/n = .9997000899730081 * c Air propagation center is dragged v*(1-(1/n^2)) = .2398920431837794 m/sec east of the ECI frame. Beam source is 400 - .2398920431837794 = 399.7601079568162 m/sec further east than the propagation center, relative to the speed of light with no medium. Relative to the speed of light in air, the propagation base is 399.7601079568162 / .9997000899730081 = 399.8800359892033 m/sec to the west. The distance that the beam must travel along the air path increases pointing east and decreases pointing west, at the rate of 399.8800359892033 meters per second of path length. Number of wavelengths along the path if the device is fixed with the ECI frame = pl/(wl*1/n) = 2839968.454258675 Number of waves along the path with the device moving east at 400 m/sec and the beam pointing east = (((c+((v-(v*(1-(1/n^2))))/(1/n)))/c)*pl)/(wl*1/n) = 2839972.239747634 which is 3.785488959401846 more waves than the ECI frame fixed device. Number of waves along the path with the beam pointing west = (((c-((v-(v*(1-(1/n^2))))/(1/n)))/c)*pl)/(wl*1/n) = 2839964.668769716 which is 3.785488958936185 less waves than the ECI frame fixed device. That's exactly the same as for the water path.
No matter what mediums are used, and how many waves are along each
path, exactly the same number of wavelengths will be added to or
subtracted from each path as the device is rotated. No interference
pattern shift could possibly be noted unless the water is flowing.
---------------
The experiment only demonstrates that light propagates on a base which is located somewhere to the west by at least the velocity of the water flow.