We report on Keck and Hubble Space Telescope observations of the binary millisecond pulsar PSR B1855+09 .
We detect its white-dwarf companion and measure m _ { F 555 W } = 25.90 \pm 0.12 and m _ { F 814 W } = 24.19 \pm 0.11 ( Vega system ) .
From the reddening-corrected color , ( m _ { F 555 W } - m _ { F 814 W } ) _ { 0 } = 1.06 \pm 0.21 , we infer a temperature T _ { eff } = 4800 \pm 800 K. The white-dwarf mass is known accurately from measurements of the Shapiro delay of the pulsar signal , M _ { C } = 0.258 ^ { +0.028 } _ { -0.016 } M _ { \odot } .
Hence , given a cooling model , one can use the measured temperature to determine the cooling age .
The main uncertainty in the cooling models for such low-mass white dwarfs is the amount of residual nuclear burning , which is set by the thickness of the hydrogen layer surrounding the helium core .
From the properties of similar systems , it has been inferred that helium white dwarfs form with thick hydrogen layers , with mass \simgt 3 \times 10 ^ { -3 } M _ { \odot } , which leads to significant additional heating .
This is consistent with expectations from simple evolutionary models of the preceding binary evolution .
For PSR B1855+09 , though , such models lead to a cooling age of \sim 10 Gyr , which is twice the spin-down age of the pulsar .
It could be that the spin-down age were incorrect , which would call the standard vacuum dipole braking model into question .
For two other pulsar companions , however , ages well over 10 Gyr are inferred , indicating that the problem may lie with the cooling models .
There is no age discrepancy for models in which the white dwarfs are formed with thinner hydrogen layers ( \simlt 3 \times 10 ^ { -4 } M _ { \odot } ) .