It seems to me that a monolayer of cells on the bottom surface of the well, or a suspension of cells, should scatter light to some measurable degree since they can both be perceived with the eye. The signal might be quite small, however. It could be maximized by using the shortest possible wavelength. In a polystyrene plate, this is probably about 340 nm.

Another issue is that there is a small variation between wells in the absorbance signal even when the plate is empty. This variation might be large relative to the cell scattering signal, ruining the measurement. Certain types of plates may be better suited to this measurement than others. I would suggest using glass-bottom plates because of the superior optical properties.

If you can find a nephelometer, a type of plate reader that measures the intensity of scattered light directly, rather than the apparent absorbance caused by light scattering, you might have a better signal.

Ultimately, you would have to calibrate the signal by also counting the number of cells directly and correlating that to the plate reader measurement.

I think you cannot. These spectrophotometer reads % of light waves that get reflected back, and most of the mammalian cells are too thin for their absorption spectrum to be measured. Indeed you mentioned you do not want to use any viability assay but you can use AlamarBlue, for which you can measure absorbance and fluorescence together.

It seems to me that a monolayer of cells on the bottom surface of the well, or a suspension of cells, should scatter light to some measurable degree since they can both be perceived with the eye. The signal might be quite small, however. It could be maximized by using the shortest possible wavelength. In a polystyrene plate, this is probably about 340 nm.

Another issue is that there is a small variation between wells in the absorbance signal even when the plate is empty. This variation might be large relative to the cell scattering signal, ruining the measurement. Certain types of plates may be better suited to this measurement than others. I would suggest using glass-bottom plates because of the superior optical properties.

If you can find a nephelometer, a type of plate reader that measures the intensity of scattered light directly, rather than the apparent absorbance caused by light scattering, you might have a better signal.

Ultimately, you would have to calibrate the signal by also counting the number of cells directly and correlating that to the plate reader measurement.

Cells are particles capable of scattering light laterally. This makes it possible to measure absorbance in both spectrophotometers and microplate photometers (elisa reader). The measurement is not a measure of actual absorption but of turbidity because the cells attenuate the beam of light with which they are irradiated by reflecting or refracting part of this light. In a suitable cell concentration range, the apparent absorbances of the instrument have a proportional relationship with the cell density.

In our laboratory we have experience discriminating cell concentrations in bacterial cultures for example. In fact, we can monitor cell growth for several hours in the presence of substances with antimicrobial properties.

You do not say it explicitly in your approach, but I suppose in your case it's about human cells. If so, it is necessary that your cells are in suspension and you must ensure that they do not form clusters, that is, before measuring the absorbances you must agitate them so that they separate into individual particles as much as possible. This you must do before placing them on the plate and just before measuring the absorbance I recommend using the agitation function that the reader has to shake the plate. All this agitation is necessary because the cells behave like unstable colloidal systems, the cells settle, come together again, and this affects the reproducibility of the scattering of light. Agitation is then particularly critical if you later want to establish significant differences using statistical tests.

Finally, these apparent absorbances can not be taken as a measurement of vaibility, this measurement is not able to discriminate viable cells from dead cells.

I think it is quite possible to use a spectrophotometer to detect a difference in the cell density within different wells (of 6 or 96 well plate) using only media.

The cells in suspension (e.g., yeast or bacteria) it is standard to use Optical Density (OD, based on light scattering) to measure the cell concentration. This will work in any transmission spectroscopy system (cuvettes or 96-well plate). It is customary to pick a visible wavelength to make the measurement and for good work the Optical Density (AU at lambda) should be correlated to cell mass (g dry mass per unit volume). Visible wavelenghts (e.g., 640 nm) are used because most cell components don't absorb there. All bets are off if your cells have chloroplasts and you have to work around their intrinsic absorption. This is still possible with cells that grow in biofilms or layers (e.g., mammalian cells), but is a bit trickier. Since biofilms don't have the same reflective curvature and surfaces to scatter the light, you have to base the measurements of the internal subcellular organelles (e.g., nucleus) or total protein at a tyrosine or tryptophan absorbance 215-235 nm. Again, good practice is to correlate this OD measurement to dry mass or cell numbers. The key issue is that as growth conditions change and the cell size or shape changes, the OD measurement will change relative to other measures of cell mass or abundance, so you need to cross correlate these measrements at each condition.