Not to pick on Gary, but a couple of comments need to be made here. This particular field has a terrible sequence, especially when the Mira is bright. The brightest comp is star 3, with V=9; the next brightest comp is 3 magnitudes fainter. When you are working with a variable at V=7, you have too wide a dynamic range necessary to get good signal/noise on more than the brightest comparison star. This gets even worse when you look at the B magnitudes of the comp stars and of the variable. So first, Gary should *only* be using star 3 in his reductions when the variable is this bright. This is probably the cause of most of Gary's problems (other than the mentioned typos, etc.). Miras in general are a poor example of whether CCD observers are more or less acccurate than visual observers. These stars have very red colors and lots of molecular bands. You cannot transform them properly and different filters from different vendors, especially where molecular bands come into play (Ic), will be difficult to match between observers. CCD V-band measures will differ markedly from visual measures due to the redness and different response between the Johnson V filter and the human eye. Finally, Gary's comment that these are unevaluated (raw) measures is entirely appropriate. I bet that you can find wide variations in photometry for a given star, visual or otherwise, reported to any variable star organization; it is not just the province of the AAVSO. What is usually missing is the feedback from the data collecting group to the observers to let them know that their data has problems. This is not easy for photometry, as seemingly smoothly varying stars can often have bumps and wiggles that are real yet look like an incorrect measure (for example, an eclipse that is rarely seen). Sebastian has been useful in this regard for U Ori; I doubt he is willing to spend the time to do this for a few thousand more stars! You have to leave all the warts in the raw data until some later time when comparison with other photometry might help in validation. The quick-look plotting facility of the AAVSO site is marvelously useful for the individual observer in comparing his/her data to others. For minor planet astrometric work, there is an easy computerized feedback mechanism since the errors of each observation with respect to the current astrometric orbit can be calculated and returned to the observer; it is rare that an orbiting body will suddenly jump by thousands of kilometers from its predicted path! The real bottom line is that it is up to the individual photometrist to look at his/her data, compared with others, and decide whether it seems reasonable or has possible error. If the latter, then you need to do what Gary did: see if there is an obvious reason for the potential error (and correct it if so), or if no obvious reason, leave it in the database for analysts to decide for themselves. If corrections are necessary, the database maintainer has to be able to replace the incorrect observation and not leave it in the file. As a sidebar, I should note that if you find that you are making many errors, you should look at the seemingly ok data as well as the deviant points since these could also have errors that just happen to balance out. Arne > > The next observation in your email was about an observation done 9 days later > by (WGR), which is myself, at V = 6.97 and B-V of 3.69---actually I reported > B of 10.66. This does seem unusual. I went back and looked at the > photometry sheets for these two. I did find that my chart showed comp star 5 > with a magnitude of 12.991, while the latest chart on the web shows 13.991. > I also had a note on my chart that Ron Zissell had found this error, but I > used the old value. I have corrected my chart. This would change my value > to V= 7.31. I gave the B observation two different days of scrutiny, and > could find nothing obvious. The comps are in the right places and are the > same ones used in the V observation. The 10.66 is the average of 3 > observations which gave 10.18, 10.73, and 11.09 for comps 5, 4 and 3. This > gives a Std Dev of .37 mag (I know that with 3 samples the Std Dev is not > appropriate, but I use it to measure the scatter of my observations on every > one). Good numbers may be as los as .03 mag, and rarely better. So No > Closure on the B observation.