A simple field trial was performed by students from the McMaster University Shad Valley summer program. Each observer was given an observing tube with a length of 16cm and a radius of 2cm.
Each set of observations was taken by students working in pairs, one partner to count stars and the second student to act as data recorder.
The results are shown in the histogram below.
In a post-observation debriefing the following facts were discovered.
Not all participants allowed sufficient time to allow for full night-vision adaptation to occur. (Dark adaptation takes at least 20 minutes.)
For those that allowed plenty of time for night vision adaptation to occur, a few had their adaptation compromised by the use of flashlights during the observation period. (Loss of dark adaptation takes only a few seconds.)
Observations taken with the Moon visible resulted in lower star counts than observers who used a tree to building to "hide" the Moon from view. (Ideally, observation should only be taken with no Moon above horizon.)
Observers who counted stars using a slightly "averted" view through the tube counted many more stars than observers who looked directly at the star field through the tube. (Averted vision is much more sensitive to low light levels than direct vision.)
The average number of stars visible in the sky (for all observers) is calculated to be 114 stars with a standard deviation of 47 (very large!).
This large standard deviation is a result of procedural inconsistencies i.e. failure to rigorously follow the protocol.
A series of observations were made under nearly ideal observing conditions, by a one observer ( to avoid physiological bias) using the same technique (to avoid procedural bias).
The observations were made at approximately (local) midnight on July 17th, under perfectly clear skies with no local lighting and an unobscured horizon.
In this experiment a tube with a diameter of 4cm (r=2) was used to make a statistical star count, varying only the tube length.
Multiple observations were taken at with the tube length of 28, 22, 16 and 12 centimetres.
The results are shown below.
Based on this simple experiment it would appear that a tube_length/tube_radius ratio of about 10:1 is optimum for most sky conditions.