The accuracy does not matter, when making a relative comparison. As long as the response of the meter is linear, a relative measurement will always yield the correct percentage.
The problem with this is that it is hard to tell how much power is coming from the diode, if it is in an AixiZ module. Part of the light cone hits the inner walls before it can escape.
And then, the measurement can be influenced by the distance of the diode.. The closer you bring it, the more you measure, even tho the entire light was covering the sensor from the start. This is because the thermopile can measure the heat coming from the diode in the form of IR..
Otherwise, a decrease of 11% would mean, there is 13% room for improvement. The acrylic does not decrease power by 30% in comparison to the Meredith. The Meredith increases the power by 30% in comparison to the acrylic.
The Meredith would seem to be one hell of a lens. If this lens comes out similar, i'll be happy with it. Regardless of the actual power, it is still a lens designed for 405nm, so it should be good at it..
And if, for some strange reason, it should turn out not to be, nobody loses anything, except for me. So let's just treat this as an experiment for now..