Why are the 'Fixed Stars' so FIXED?

On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:

....
Have you seen this?http://mb-soft.com/public2/cepheid.html

He cites "Duncan" without further information. A bit
of a browse suggests it is a curve from 1921 !

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

George

On 13 Mar 2007 01:35:25 -0700, "George Dishman"
wrote:

On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:

...
Have you seen this?http://mb-soft.com/public2/cepheid.html

He cites "Duncan" without further information. A bit
of a browse suggests it is a curve from 1921 !

I agree the whole paper was very suspect.

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

Yes I saw that one.
It clearly shows a first overtone oscillation, not present in the Duncan curve.
Mind you, the four cycles shown in this curve are just repeats of the first
cycle....so you don't really know how accurate it is, either.

The frequent presence of harmonics in star curves caused me to tentatively
accept the idea of huff-puff stars. It is pretty hard to explain the phenomenon
purely on orbital grounds. There wouldn't be all that many stars that have
satelites orbiting with periods in the exact ratio of 2:1. Maybe there are
'figure of eight' orbits.
Anyway, the BaTh still applies to huff-puff stars, just as it would for
orbiting ones.

You can add and see the effects of a first overtone in my original
variables.exe program.

George


"When a true genius appears in the world, you may know
him by this sign, that the dunces are all in confederacy against him."
--Jonathan Swift.

On 13 Mar 2007 01:24:19 -0700, "George Dishman"
wrote:

On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:
On 12 Mar 2007 05:11:04 -0700, "George Dishman" wrote:

Right, it is just a step along the way. As you say
later there should be a statistical spread and if
you predicted every pulsar orbit had to be within
a few mas of face-on, it would indicate a problem.

The problem remains to establish a figure for the true orbital velocity given
that all we have available is the willusion.

Sure, I'm working towards explaining some of the
methods you can use to do that.

As far as I can see, the velocity curve calculated 'classically' from observed
doppler shifts should be exactly the same as my brightness curve,
The phase difference between the blue and green curves appears to remain the
same 90deg for distances of 0.2LYs and 200LYs.

However, I was
going to point out that a face-on conflicts with
the observation of a Shapiro delay so we can
discard that idea.

But the phase of the supposed Shapiro delay is based on standard Doppler.

No, the Shapiro delay must peak at the time of
superior conjunction when the line of sight
passes closest to the dwarf.

....but even Shapiro delay has a different meaning in the BaTh....although the
phase of maximum effect should still coincide with the dwarf being furthest
away. The 'delay' should be negative when the dwarf is closest.

Sort of, you are right that the 'delay' should be
negative but the effect should still be most
significant when it is farthest. As a deviation
from the normal Keplerian effects we usually discuss
GR gives this change of arrival time as a function of
phase:

Advance
|______ ______|
| \/ |
Delay

while ballistic theory predicts this:

Advance
|______/\______|
| |
Delay

Bear in mind we don't know the exact distance to
the pulsar to a few metres, it is only a relative
shift on top of the orbital effects and also the
proper motion of the whole system.

Laying aside the inversion, it still gives a valid
phase reference.

Well I don't even accept that the effect currently attributed to 'Shapiro
delay' IS actually just that. According to the BaTh, there will be a slowing of
light as it escapes the gravitational influence of the pair. The amount depends
somewhat on the position in the orbit where the light was emitted.
The final speed will not vary by much but the travel time to a distant observer
might be significantly affected. Maybe I should investigate this further...and
maybe include it in my program.
Again, it would help enormously if we could obtain brightness and velocity
curves for the dwarf.

I reckon what is happening in the case of this Pulsar is that the very heavy
neutron star is wobbling relatively slowly around its barycentre with the
dwarf.

Trouble is that you cannot have a "very heavy neutron
star", there is a limit to the ability of neutrons to
resist being crushed.

...theories, theories......

Its orbital speed could easily be less than 1 km/s. The pitch angle
might be around 60 degres or less.

The pitch can be found from the Shapiro delay just by
comparison with the empirical delay measured near the
Sun without worrying about any particular theory.

George, the situation is that there is an apparent anomaly in the pulse arrival
rate that appears to be explainable by the shapiro effect.
I'm not prepared to accept that this is the correct or only explanation. The
BaTh opens up other possibilies. You only have to look at the so called
'eclipsing binary' curves to realise that. I would say that most of these
'eclipse-like' curves are just a result of c+v, where the orbits are moderately
eccentric and the periastron is nearest to the observer. One can only tell the
difference if accurate spectral data is available.

Have a look yourself. Set Yaw angle at -90, eccentricity at ~0.7.
There is even evidence of the second small dip.

What would help greatly would be a brightness curve of the dwarf...or at least
something about its velocity variations.

I haven't seen one yet but since I don't accept
your "incompressible photon" idea it probably
wouldn't help. I want to see how far we can get
using _only_ the pulses where we agree the effects.

Well I don't think we can go much further.
All I can derive is the product (extinction distance x true velocity)

That is controlled
solely by the geometry of the orbit so it
provides a _reference_ against which the phase
of the Doppler can be measured. Conventionally
for a circular orbit it would be at a point of
zero shift but for ballistic theory the Doppler
will be a combination of velocity and acceleration
terms so the phase relative to the reference can
tell you the ratio of the velocity and acceleration
contributions.

I suggest we use the terms ADoppler and VDoppler to distinguish between these
two. ADoppler includes a VDoppler component.

OK, but exclude the VDoppler from the ADoppler and
call the combination the total.

OK, TDoppler.
The problem is that my original method (counting pulses that arrive in a
certain time interval) should take both effects into account...and its curves
are identical to those produced with 'george'.

How do we arrive at a true velocity curve when all we have is the ADoppler
willusion? Because of hte VDoppler component, the velocity curve will not
normally be quite the same as my 'brightness' curve. It will be out of phase
and have a different shape until it stabilizes with distance.

That's the key, the phase depends on the ratio so
given the Shapiro marker we should be able to find
a maximum value for the extinction distance.

As far as I can see, there is no change in phase with distance for circular
orbits. It remains at 90.
There appears to be a small change in the case of highly eliptical orbits.


If we use a computer simulation we cannot assume the hipparcos distance is
applicable unless we also assume zero extinction....and that is not something I
would like to do at this stage.

No, what we do is use a combination, if you set the
program distance to the Hipparcos level, you are
assuming no extinction and the result will usually
be multiple images which is ruled out. If you then
set it much less, you are assuming an observe at
infinity and the distance is the extinction. Having
done both, as long as the ratio is large (i.e. the
observer is at a much greater distance than the
extinction) then you can use the latter method. If
it turns out the distances are similar, then you
have to work out the exponential to get a more
accurate figure for extinction using the known
Hipparcos range.

Yes. Generally, the distance I select will be the 'extinction distance'.

...you seem to be having trouble riding your mind of everything you have been
taught in the past George.

You seem to have trouble realising that phase
tells you lots about the situation :-)

No, I can see that the V component will shift the red curve away from the
'brightness curve'...but we need the extinction distance before we can go much
further.

For a circular orbit, when the phase is 45 degrees
relative to the Shapiro peak, the ADoppler component
is equal to the VDoppler which tells you the extinction.
It will be more complex for an elliptical orbit but
that's where you program comes in.

I think the V component is always going to be much smaller than the A.
I can only detect very small phase changes even for highly eccentric orbits.

I have upgraded my program again so that you can see any number of orbits and
can increase the y scale by any factor.

If you click anywhere on the screen showing the curves a vertical line comes up
so you can compare phases.
Note, the brightness curve increases upwards, the velocity curve increases down
the screen.

Have you seen this?http://mb-soft.com/public2/cepheid.html

Crank crap I'm afraid. He makes the point that the
infalling acceleration is only 0.16 m/s^2 while the
surface gravity is 0.93 m/s^2 and suggests that's a
problem. Of course all it means is that the upward
pressure has dropped from slightly more than 0.93
when the star was expanding to about 0.77 m/s^2 or
5/6ths of the gravity when it is collapsing.

I had a look at some of his other stuff and it is
pretty clueless.

I agree. But Jerry made a big issue of it.

... The interesting question that
we will get to soon is what extinction distance
changes the phase to 45 degrees. You might have
to think a bit to see what that will tell us ;-)

Yes I can see the point. ..ADoppler starts to dominate.

Not only that, the two are equal at that point
so you know the extinction. Below that or in
particular for smaller phase angles, the phase
becomes close to proportional so if you calculate
the extinction at say 10 degrees then that at 1
degree will be about 1/10th the distance. It lets
you work out the maximum value based on the
uncertainty in the phase shift of the conventional
analysis.

There is something wrong here.
For circular orbits, I get the same phase relationship for 0.1 LYs and
100LYs....exactly 90 deg.

I think we have to incorporate the rate of change of 'bunching' rather than
just 'time between pulses'.

George


"When a true genius appears in the world, you may know
him by this sign, that the dunces are all in confederacy against him."
--Jonathan Swift.

"Henri Wilson" HW@.... wrote in message
news
On 13 Mar 2007 01:35:25 -0700, "George Dishman"
wrote:

On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:

...
Have you seen this?http://mb-soft.com/public2/cepheid.html

He cites "Duncan" without further information. A bit
of a browse suggests it is a curve from 1921 !

I agree the whole paper was very suspect.

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

Yes I saw that one.
It clearly shows a first overtone oscillation, not present in the Duncan
curve.

By eye, the curve is more complex than just a single overtone.

Mind you, the four cycles shown in this curve are just repeats of the
first
cycle....so you don't really know how accurate it is, either.

The obvious approach to smoothing and combining measurements
over more than one cycle would be to take a Fourier transform,
identifiy the fundamental, set all the bins other than that
and the harmonics to zero and reverse transform. The pattern
would then repeat perfectly of course but would contain all
the original information folded into each cycle. It might be
worht contacting the author, if he gave you the amplitude and
phase of the harmonics it would be fairly straightforward to
reconstruct the curve and calculate the residuals.

The frequent presence of harmonics in star curves caused me to tentatively
accept the idea of huff-puff stars. It is pretty hard to explain the
phenomenon
purely on orbital grounds. There wouldn't be all that many stars that have
satelites orbiting with periods in the exact ratio of 2:1. Maybe there are
'figure of eight' orbits.

There are some wierd shapes around but you would have to
simulate them to get the actual curves, they would be far
too hard toestimate given the importance of the acceleration
term.

Anyway, the BaTh still applies to huff-puff stars, just as it would for
orbiting ones.

It would apply, but the predictions would be different
because the acceleration due to expansion and contraction
would also be significant.

You can add and see the effects of a first overtone in my original
variables.exe program.

For the moment I'll let you do the driving, I find the GUI
somewhat unfriendly.

George

Just a quick note on one point, I'll reply in more
detail tomorrow.

"Henri Wilson" HW@.... wrote in message
...
On 13 Mar 2007 01:24:19 -0700, "George Dishman"
wrote:
On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:
On 12 Mar 2007 05:11:04 -0700, "George Dishman"
wrote:
....
... The interesting question that
we will get to soon is what extinction distance
changes the phase to 45 degrees. You might have
to think a bit to see what that will tell us ;-)

Yes I can see the point. ..ADoppler starts to dominate.

Not only that, the two are equal at that point
so you know the extinction. Below that or in
particular for smaller phase angles, the phase
becomes close to proportional so if you calculate
the extinction at say 10 degrees then that at 1
degree will be about 1/10th the distance. It lets
you work out the maximum value based on the
uncertainty in the phase shift of the conventional
analysis.

There is something wrong here.
For circular orbits, I get the same phase relationship for 0.1 LYs and
100LYs....exactly 90 deg.

Sure, but try it for an extinction of a few light
hours, or maybe light minutes. That's the point
Henry, the pulsar will give much tighter limits
than you have been used to in the past.

I think we have to incorporate the rate of change of 'bunching' rather
than
just 'time between pulses'.

No, barring bugs your program should give us what
we want now but you might have to handle smaller
scales than you expected. There are some subtleties
around the Shapiro effect but I think they will
prove to be inconsequential.

George

On 13 Mar, 23:35, HW@....(Henri Wilson) wrote:
On 13 Mar 2007 01:24:19 -0700, "George Dishman" wrote:
On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:
On 12 Mar 2007 05:11:04 -0700, "George Dishman" wrote:
....
The problem remains to establish a figure for the true orbital velocity given
that all we have available is the willusion.

Sure, I'm working towards explaining some of the
methods you can use to do that.

As far as I can see, the velocity curve calculated 'classically' from observed
doppler shifts should be exactly the same as my brightness curve,

Yes, other than at high values.

The phase difference between the blue and green curves appears to remain the
same 90deg for distances of 0.2LYs and 200LYs.

Our previous estimate was about 6 light hours so
at 0.2 light years the acceleration dominates
completely.

....but even Shapiro delay has a different meaning in the BaTh....although the
phase of maximum effect should still coincide with the dwarf being furthest
away. The 'delay' should be negative when the dwarf is closest.

Sort of, you are right that the 'delay' should be
negative but the effect should still be most
significant when it is farthest. As a deviation
from the normal Keplerian effects we usually discuss
GR gives this change of arrival time as a function of
phase:

Advance
|______ ______|
| \/ |
Delay

while ballistic theory predicts this:

Advance
|______/\______|
| |
Delay

Bear in mind we don't know the exact distance to
the pulsar to a few metres, it is only a relative
shift on top of the orbital effects and also the
proper motion of the whole system.

Laying aside the inversion, it still gives a valid
phase reference.

Well I don't even accept that the effect currently attributed to 'Shapiro
delay' IS actually just that.

It has exactly the form that is expected on purely
empirical grounds, we see the effect locally and an
identical effect for the pulsar. That makes no
assumption about its cause merely recognises its
existence.

Whatever it is, it still tells you the phase which
is all we are using it for.

According to the BaTh, there will be a slowing of
light as it escapes the gravitational influence of the pair.

At some distance from the system there will be a
slowing which is close to the effect of a point
mass, the separation of the two bodies becomes
negligible. The Shapiro delay is how the slowing
varies relative to that mean effect as a function
of the phase.

The amount depends
somewhat on the position in the orbit where the light was emitted.

It is that part only that interests us.

The final speed will not vary by much but the travel time to a distant observer
might be significantly affected. Maybe I should investigate this further...and
maybe include it in my program.

To start with I suggest you write a separate program
just to look at that effect. You need to sort out
getting a delay instead of an advance first and that
will be a problem, ballistic theory gets it wrong.

Again, it would help enormously if we could obtain brightness and velocity
curves for the dwarf.

I reckon what is happening in the case of this Pulsar is that the very heavy
neutron star is wobbling relatively slowly around its barycentre with the
dwarf.

Trouble is that you cannot have a "very heavy neutron
star", there is a limit to the ability of neutrons to
resist being crushed.

..theories, theories......

That's all science is Henry, mathematical equations
proven by observation and called 'theories'. In this
case the observations are from high energy particle
physics. You can argue there is some doubt about the
upper limit but a factor of 2 is probably as much as
is credible and a factor of 10 certainly excessive.

Its orbital speed could easily be less than 1 km/s. The pitch angle
might be around 60 degres or less.

The pitch can be found from the Shapiro delay just by
comparison with the empirical delay measured near the
Sun without worrying about any particular theory.

George, the situation is that there is an apparent anomaly in the pulse arrival
rate that appears to be explainable by the shapiro effect.

... that exactly matches the measured Shapiro effect for
the Sun.

I'm not prepared to accept that this is the correct or only explanation. The
BaTh opens up other possibilies. You only have to look at the so called
'eclipsing binary' curves to realise that. I would say that most of these
'eclipse-like' curves are just a result of c+v, where the orbits are moderately
eccentric and the periastron is nearest to the observer. One can only tell the
difference if accurate spectral data is available.

Have a look yourself. Set Yaw angle at -90, eccentricity at ~0.7.
There is even evidence of the second small dip.

I doubt you will get those results now you are
taking acceleration into account correctly but
as you say the spectral data is key. I'm not
going to start looking at that, it's a quite
different discussion.

What would help greatly would be a brightness curve of the dwarf...or at least
something about its velocity variations.

I haven't seen one yet but since I don't accept
your "incompressible photon" idea it probably
wouldn't help. I want to see how far we can get
using _only_ the pulses where we agree the effects.

Well I don't think we can go much further.

You're always eager to give up Henry, what are
you afraid of?

All I can derive is the product (extinction distance x true velocity)

Not true, the phase data will tell you much more
and the harmonic content of the velocity curves
more again.

....
How do we arrive at a true velocity curve when all we have is the ADoppler
willusion? Because of hte VDoppler component, the velocity curve will not
normally be quite the same as my 'brightness' curve. It will be out of phase
and have a different shape until it stabilizes with distance.

That's the key, the phase depends on the ratio so
given the Shapiro marker we should be able to find
a maximum value for the extinction distance.

As far as I can see, there is no change in phase with distance for circular
orbits. It remains at 90.

If you set the extinction to zero distance, you must
recover the conventional result which has only the
VDoppler. If you are getting a 90 degree shift then
the ADoppler is dominating and your extinction is too
high. You see Henry, there is still more to learn.
Previously you agreed the extinction must be less than
a light day and my estimate was 6 hours so in good
agreement, and that was just an upper limit. You need
to adjust your program so that we can investigate
numbers in the range of light hours and light minutes
would probably be better.

There appears to be a small change in the case of highly eliptical orbits.

Possibly, but I would want to see a harmonic analysis
to see if you could get a match. If you can't build
a Fourier transform into your program, add an export
button to write a csv file and analyse it using Excel.

If we use a computer simulation we cannot assume the hipparcos distance is
applicable unless we also assume zero extinction....and that is not something I
would like to do at this stage.

No, what we do is use a combination, if you set the
program distance to the Hipparcos level, you are
assuming no extinction and the result will usually
be multiple images which is ruled out. If you then
set it much less, you are assuming an observe at
infinity and the distance is the extinction. Having
done both, as long as the ratio is large (i.e. the
observer is at a much greater distance than the
extinction) then you can use the latter method. If
it turns out the distances are similar, then you
have to work out the exponential to get a more
accurate figure for extinction using the known
Hipparcos range.

Yes. Generally, the distance I select will be the 'extinction distance'.

Yes, once you do the basic check, but for brightness
changes of less than 2 or 3 mag. that will be how it
works.

No, I can see that the V component will shift the red curve away from the
'brightness curve'...but we need the extinction distance before we can go much
further.

For a circular orbit, when the phase is 45 degrees
relative to the Shapiro peak, the ADoppler component
is equal to the VDoppler which tells you the extinction.
It will be more complex for an elliptical orbit but
that's where you program comes in.

I think the V component is always going to be much smaller than the A.
I can only detect very small phase changes even for highly eccentric orbits.

That's good, it should mean we can separate extinction
and eccentricity effects. For the pulsar the phase is
identical to the conventional prediction for an
eccentricity of 10^-7 so even a high eccentricity with
your program should produce quite a small phase change
and that means the VDoppler must be much greater than
the ADoppler. That in turn means a very small extinction
distance.

I have upgraded my program again so that you can see any number of orbits and
can increase the y scale by any factor.

If you click anywhere on the screen showing the curves a vertical line comes up
so you can compare phases.
Note, the brightness curve increases upwards, the velocity curve increases down
the screen.

I'll leave you to do the driving and look forward to
seeing some screen shots. What I want to see first is
an edge-on circular orbit with a (roughly) 45 degree
phase shift and a red curve of 27983 m/s peak. To get
that you will need a distance of less than a light day.

The next step after that will be to try to get a real
match using your program but let's see what that rough
estimate is first.

Not only that, the two are equal at that point
so you know the extinction. Below that or in
particular for smaller phase angles, the phase
becomes close to proportional so if you calculate
the extinction at say 10 degrees then that at 1
degree will be about 1/10th the distance. It lets
you work out the maximum value based on the
uncertainty in the phase shift of the conventional
analysis.

There is something wrong here.
For circular orbits, I get the same phase relationship for 0.1 LYs and
100LYs....exactly 90 deg.

As expected, you need to change the program to cope
with _much_ smaller distances.

George

On Mar 13, 2:35 am, "George Dishman" wrote:
On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:

...

Have you seen this?http://mb-soft.com/public2/cepheid.html

He cites "Duncan" without further information. A bit
of a browse suggests it is a curve from 1921 !

Actually the data is from 1908
http://articles.adsabs.harvard.edu/c...icOB...5...82D

If the above link is truncated, you can search for the paper:
"The orbits of the Cepheid variables Y Sagittarii and RT Aurigae :
with a discussion of the possible causes of this type of stellar
variation", by John Charles Duncan, 1908

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

Yes, instrumentation has advanced a wee bit in the last century...

Here is a 1969 publication by Bappu et al.
http://adsabs.harvard.edu/abs/1969MNRAS.142..295B

Jerry

On 14 Mar, 12:46, "Jerry" wrote:
On Mar 13, 2:35 am, "George Dishman" wrote:
On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:
Actually the data is from 1908

http://articles.adsabs.harvard.edu/c..._query?1908Lic...

If the above link is truncated, you can search for the paper:
"The orbits of the Cepheid variables Y Sagittarii and RT Aurigae :
with a discussion of the possible causes of this type of stellar
variation", by John Charles Duncan, 1908

A real bit of history then.

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

Yes, instrumentation has advanced a wee bit in the last century...

Here is a 1969 publication by Bappu et al.http://adsabs.harvard.edu/abs/1969MNRAS.142..295B

Excellent, thanks. Given the importance of Cepheids,
I'm a surprised there aren't more recent studies, or
is that just the latest you know of?

Thanks for the data Jerry, it will be intersting to
see what numbers Henry can come up with to fit those,
though it is already clear he will get a poor fit at
best.

George

"Jerry" wrote in message oups.com...
On Mar 13, 2:35 am, "George Dishman" wrote:
On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:

...

Have you seen this?http://mb-soft.com/public2/cepheid.html

He cites "Duncan" without further information. A bit
of a browse suggests it is a curve from 1921 !

Actually the data is from 1908
http://articles.adsabs.harvard.edu/c...cOB....5...82D

If the above link is truncated, you can search for the paper:
"The orbits of the Cepheid variables Y Sagittarii and RT Aurigae :
with a discussion of the possible causes of this type of stellar
variation", by John Charles Duncan, 1908

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

Yes, instrumentation has advanced a wee bit in the last century...

Here is a 1969 publication by Bappu et al.
http://adsabs.harvard.edu/abs/1969MNRAS.142..295B

Jerry


What more evidence do you need for the PoR?

On Mar 14, 8:47 am, "George Dishman" wrote:
On 14 Mar, 12:46, "Jerry" wrote:

On Mar 13, 2:35 am, "George Dishman" wrote:
On 12 Mar, 22:11, HW@....(Henri Wilson) wrote:
Actually the data is from 1908

http://articles.adsabs.harvard.edu/c..._query?1908Lic...

If the above link is truncated, you can search for the paper:
"The orbits of the Cepheid variables Y Sagittarii and RT Aurigae :
with a discussion of the possible causes of this type of stellar
variation", by John Charles Duncan, 1908

A real bit of history then.

This is a more recent curve from an amateur:

http://www.student.oulu.fi/~ktikkane/AST/RTAUR.html

There is far more detail in that which shows how far
the capabilities have come, and I suspect you'll get
even better data if you do some serious searching.
Note the amount of detail in the variations.

Yes, instrumentation has advanced a wee bit in the last century...

Here is a 1969 publication by Bappu et al.
http://adsabs.harvard.edu/abs/1969MNRAS.142..295B

Excellent, thanks. Given the importance of Cepheids,
I'm a surprised there aren't more recent studies, or
is that just the latest you know of?

A little googling found this excellent reference:
Physical properties of the Cepheids RT Aurigae and SZ Tauri
Authors: Gieren, W. P.
Astronomy and Astrophysics (ISSN 0004-6361),
vol. 148, no. 1, July 1985, p. 138-144.

http://articles.adsabs.harvard.edu//...00141.000.html

I also found several additional likely references in subscriber-
access journals. I'll check on them the next time that I visit
the main campus library.

Thanks for the data Jerry, it will be intersting to
see what numbers Henry can come up with to fit those,
though it is already clear he will get a poor fit at
best.

By futzing around with his parameters, Henri can "sort of" match
the shape of the RT Aur radial velocity curve (to the point
where Henri will claim a decent match, but no objective observer
would agree), but even Henri will admit that he can't match the
phasing relative to the luminosity curve.

Jerry