Powering up a photomultiplier (PMT)

I want to get a photomultiplier working. No particular use yet, just to get a meaningful reading off it.
The device in question: http://www.melz-evp.ru/FEU85.html
Connected it as per original datasheet: ../imgqa/eboard/Antenna/rf-hujyh223oen.gif
R is 100KΩ.

For sensing, there is a 1MΩ resistor before the anode, and a voltmeter across it.
The power supply gives 900V.

Now, the problem is - there is a constant current across the resistor, starting from 90μA, and slowly going down to 60-50μA when powered, climbing back if left unpowered. Maybe further, didn't wait long.
That current is independent of the amount of light coming in.
I tried to get it as dark as possible, knowing the legendary light sensitivity of the device, but there is no change.
The thing is in a black bag in a box, the room is dark, not a single LED left on, night outside - and the current is still the same.
Trying to add more light does not change anything - all the way to leaving it naked in a dimply lit room.
There is no reaction at any reasonable light or darkness.

The question is - what can i be doing wrong?

Are there any guides on how to use a PMT?
Google is not helping, and a few DIY descriptions i found just list what is used without why.
What do the R in the voltage divider define? Would the sensitivity be less or more if i increase or decrease the R?
Can the thing be so sensitive as to be whited out in complete darkness, or is something wrong?

Still looking for guides on how to use a PMT.
I'm also interested if there is a way to check if the tube is alive or not.

Can anyone help out?

I worked with electron multiply before and it's kind of similar to PMT. You know there is a resistor ladder across the HV? The resistors are used to bias each of the dynode. The resistor chain do draw current regardless whether you have any photon hitting or not. You said when it first turn on, it draws 90uA....say 100uA. For 1KV, that implies the total resistance is 10MΩ. Sounds a little bit low, but if you said it drops down to 50uA, that would be 20MΩ. That is not too unreasonable. Look at the schematic you provided, each R is around 1MΩ which is reasonable. I am not sure there is anything wrong.

I am not specialize in multipliers, I designed receiver circuit to read the electron pulses from the multiplier. That's the extend I worked with them.