LED circuit

On Monday we built a circuit to light two LEDs (light emitting diodes), a red (LED1) and a yellow (LED2), maximally without burning them out with too much wattage.
The red LED's given voltage and current ratings were 5V and 22.75mA respectively, yellow's were 2V and 20mA.

The LEDs were placed in a circuit in parallel with a resistor (R1 and R2) in series on each. A 9V supply was given by a power supply.
Ohm's Law was used to calculate the equivalent resistances of the LEDs.
RLED1 = 219.8Ω, RLED2 = 100Ω

Using the current  and voltage ratings, theoretical voltages and resistances values were found for R1 and R2.
V1 = 4V, V2 = 7V
R1 = 176Ω, R2 = 350Ω

We only had access to 100, 150, 220, 360, and 470-Ohm resistors, so we approximated R1 with a 150Ω and R2 with a 360Ω. Their measured values were 148.2Ω and 360Ω respectively. Both were quarter-watt resistors.

Three configurations were used:

1. Both LEDs in the circuit
2. Without LED2
3. Without LED1

These are the results:

Configuration	I of LED1 (mA)	V of LED1 (V)	I of LED2 (mA)	V of LED2 (V)	I of supply (mA)
1	14.5	6.68	19.71	19.71	35.1
2	14.8	6.74	NA	NA	15.1
3	NA	NA	20.2	1.65	20.3

Using this data, I calculated a few things.
a) If the supply was just a 9V battery with a 0.2 A-hr capacity of useful voltage, the circuit in config1 would last 4.68 hours.
b) The % error between experimental and desired values of LED current were 36.3% for LED1 and 1.45% for LED2. This was caused by resistor restrictions.
c) The efficiency of the circuit in configuration 1 was found to be 40.8%, which is pretty poor.
d) Assuming current is the same, if a 6V battery was used instead of a 9V battery, ... I really don't understand this problem.

Oh, yeah, and we blew up a red LED by overloading it with voltage.