Here's the abstract:
The CPS 2000 myth states that the blaster was banned because it burst someone's eye. This study gives "globe rupture", which is the technical term for bursting an eye, a probability of zero. The rate of hyphema (e.g., a bloody eye) is not trivial, though, and is worth investigating further.Purpose: Interactive water displays are becoming increasingly popular and can result in direct eye contact.
Therefore, the purpose of this study is to investigate eye injury risk from high speed water stream impacts and
to provide biomechanically based design parameters for water toys and water park fountains.
Methods: An experimental matrix of 38 tests was developed to impact eight porcine eyes with water streams
using a customized pressure system. Two stream diameters (3.2 mm and 6.4 mm) were tested at water veloci-
ties between 3.0 m/s and 8.5 m/s. Intraocular pressure was measured with a small pressure sensor inserted
through the optic nerve and used to determine the injury risk for hyphema, lens dislocation, retinal damage,
and globe rupture for each impact.
Results: Experimental water stream impacts created a range of intraocular pressures between 3156 mmHg
and 7006 mmHg (61 psi to 135 psi). Injury risk varied between 4.4%–27.8% for hyphema, 0.0%–3.0% for lens
dislocation, and 0.1%–3.3% for retinal damage. All tests resulted in 0.0% injury risk for globe rupture. The two
water stream diameters did not result in significantly different water stream velocities (P = 0.32); however, the
variation in water stream diameter did result in significantly different intraocular pressures (P = 0.03) with
higher pressures for the 6.4 mm stream.
Conclusions: This is the first study to experimentally measure intraocular pressure from high speed water
stream impacts and quantify the corresponding eye injury risk. It is recommended that toy water guns and
water park fountains use an upper threshold of 8.5 m/s for water stream velocities to minimize the risk of
serious acute eye damage from impacts.
I've written before that I was skeptical of that because (all else the same) thicker water streams should be safer than thinner ones due to the difference in surface area. There's a reason water jet cutters use thin streams. However, it appears that I was wrong, and larger streams are less safe because they increase eye pressure more. Perhaps this makes the CPS 2000 myth more plausible, as the stream from the CPS 2000 is much larger than the largest stream tested here.
Also, I've read a similar review paper from the same research group about solid objects impacting eyes, and basically, it doesn't take much to cause serious damage. Impact rated eye protection is absolutely necessary for Nerf, paintball, etc., but sunglasses are plenty adequate for water guns.