Hi,
You are right as far as resistive load is concerned.
the calculated resistance of the bulb is when it is hot and glowing.
the moment it is switched off and takenout for resistance measurement ,
it would show far less.
It depends on the ambient temperature, the resistor will probably be alright as long as the ambient temperature isn't too high. I would use a 0.6W resistor just to be on the safe side, it's always better to operate components on the low side of their maxim rating rather than just above it otherwise they can prematurely fail.
A 300hm: resistor would be perfect for operating a white LED from a 9V battery though.
[latex]V_F = 3.5V[/latex]
[latex]I_{F} = \frac{V-V_F}{R}=\frac{9-3.5}{300} = 18.33 \times 10^{-3}A[/latex]
[latex]P_{R}= I^2 R = (18.33 \times10^{-3})^2 \times 300 = 100.8 \times 10^{-3}W[/latex]
100.8mW isn't going to damage the resistor and most white LEDs can take 20mA so 18.33mA isn't going to do it any harm either.
Mark - just a little more on power ratings and ratings in general. The power rating of a resistor is often breifly stated but what is not stated is the conditions that go along with the rating. These "conditions" are often part of standards that manufacturer's or other groups have agreed to over the years. As already suggested, a statement that a resistor is rated at 1/4 watt tells us something but not nearly enough for all situations.
If you have an interest, take a look at the Caddock literature on the line of non-inductive power resistors. I looked at one that was rated at 100 watts. On closer inspection I noticed the detail or conditions on which the rating was based. For my application I had to de-rate or reduce the amount of power the resistor could handle. This same rating or de-rating process is what some designers must do for all components in a circuit. It is fair to say that many designers simply rely on experience as detailed design takes time and costs money.
Instant failure is not likely if you were to exceed published ratings however the ratings are a statement made by the mfr that if you remain within their published ratings you can expect reliable (though not perfect) performance. Again, experience or just a good guess is what guides many people in determining whether or not a rating is sufficient.