Intereference rings by light transmission? Recently I recorded UV-Vis spectra of liquids with a two-beam spectrometer (Perkin Elmer 900).  Lack of proper racks for cells of 0.5 cm optical path length let me use a stack consisting of a 0.2 cm cell and a one of 0.5 cm.  Such a stack was hold in both beams -- for convenience, the smaller ones filled with water, the later ones with either reference or later the solution of the analyte -- by the racks for cells of 1.0 cm.  The set-up was a quick resort, aiming only to narrow the region to be analyzed more in detail (and then, with a correct set-up of only one cell per optical path).
Magnification of the spectrum displays a fine "ondulated" finer structure on top, as shown in the picture below. Here, different curves represent repeated recordings of the same sample at low concentration of the analyte. The little shift along Abs between one curve and the next one may be due to decomposition of the sample.
Yet because these little waves are seen across the whole range recorded (190--800 nm), regardless of the samples analyzed, I conclude it is due to the experimental design, likely the stacks of two cells each, placed back-to-back.
This is the question:  Are these due to interference as Newton/Fringe rings are, even if the experimental set-up is in transmission (rather than in reflection mode)?

 A: Possibly. You could unwittingly have created an etalon.
With an etalon you get transmission maxima whenever $2l = n\lambda$, where $l$ is the width of the etalon and $n$ is an integer. I estimated the peak positions from your spectrum and graphed them with a linear regression to estimate the peak spacing:

and the spacing comes out as 15.5 nm. The spacing between a peak at $\lambda$ and the peak at the next larger wavelength is given by:
$$ n\lambda = (n-1)(\lambda + \Delta\lambda) $$
where $\Delta\lambda$ is the spacing, and a quick rearrangement gives:
$$ \Delta\lambda = \frac{\lambda}{n-1} $$
This would give the first peak at 501nm an $n$ of 35, which mean your etalon must have a thickness of around 10 $\mu$m. This is far smaller than the cell size, but could be the air gap between the cells.
The spacing varies with wavelength according to the equation above. I calculated the spacing assuming $n$ = 35 and graphed it against the values I measured:

Apart from one outlier the fit looks good, but I have to say it looks suspiciously good and I suspect the hand of chance here. I'd say my error in estimating the peak position is about 1 nm at best, so the correlation looks better than it should.
