"SWR increases when power is added."
SWR does not increase when all measured power is produced at the fundamental operating frequency.
"....Assume a ham operator is on 10 meters using a solid state amplifier. With the radio only, the SWR is 1.1:1; when the amplifier is turned on, the SWR jumps to 2.0:1. The amplifier is not only transmitting at 28 MHz, but is also transmitting on a second frequency of 56MHz. This is known as a ''second harmonic'' (2X the fundamental frequency of 28 MHz, transmitting at 56 MHz)."
http://forum.worldwidedx.com/viewtopic.php?p=112254&highlight=pushpull#112254
"Thus the SWR meter is reading both the reflected signal of the normal frequency and the rejected second harmonic signal. The antenna will not accept energy transmitted at 56 MHz, and returns it all back to the radio, which shows up on the meter as high SWR because the meter can not tell the difference between 28 MHz and 56 MHz. In fact, as much as 30% of the power is at 56 MHz."
for 30% of the power developed to be at a frequency of 56Mhz. the devices in the amplifier would have to be producing IMD2 levels worse than -4db. to -6db. when referenced to the the rated output of a single device. that's a switch, not an amplifier.
"This is generally due to an amplifier that is not adequately filtered."
no, IMD2 levels running -4db. - -6db. would be due to excessive drive and excessive operating voltages.
"Adding a Low-pass filter at the amplifier output is the only solution. For best results, connect the low pass filter directly to the amplifier using a barrel connector."
LMAO....
"Now this is to reply to the question of coax length. How many of you change the length of your coax to tune your antenna?"
well, he's not asking about that. he was referring to the length of feedline between the output of the transmitter and the input of the amplifier.
"As most people will find, varying the length of coax to the antenna will vary the SWR that the SWR meter is reporting."
"Actually, SWR should remain relatively constant no matter how long the coax is or where it is placed on the line (if its 5 feet down the coax from the radio or 50 feet down the coax from the radio)."
wrong. longer runs of feedline with their increased matched attenuation will show lower levels of standing waves than comparitively shorter runs.
"In most cases, the cause of inconsistent SWR meter readings is from poor SWR meter design or component aging / failure. For the SWR meter to read consistent SWR readings on the coax, the meter has to have an impedance itself of exactly 50 Ohms. Any deviation of the SWR meter's self impedance (from 50 Ohms) from poor design or component aging / error / failure will cause slightly inconsistent SWR readings when the SWR meters position on the coax or length of the coax is varied."
wrong. perfectly working measuring devices including expensive network analyzer probes will show different readings as they are placed at different sections in any line where the load and/or the line is anything but R=50, X=0. show me a half a dozen brand new swr meters and a source with a flat line and load and they'll all read the same anywhere in the line. show me the same six and a 50 ohm line with a mismatched load and i'll show you a laundry list of "swr readings" from several spots along a line that no longer exhibits its characteristic impedance of 50 ohms.
"In practice, generally you will find varying the coax length seemingly effects the SWR reading."
that's because the changing lengths shift the voltage minimum and voltage maximum points up and down the line. since swr meters measure minimum and maximum voltages from which the standing wave ratio is derived it consequently follows that for an accurate swr measurement of any line where R does not = 50 and X does not = 0 at the antenna input then the point in the line where the measurement is taken is critical and as i mentioned earlier for the least amount of line loading and most accurate indication the measurement must be taken at a point on the line where voltage is at absolute minimum.
SWR = EMax / Emin....if you're measuring swr at a section in the line where EMax and EMin is not occurring then you're not measuring swr. that's why the readings change when the length of any interconnecting cable is altered. only in a flat line where the characteristic impedance of the feedline is exactly equal to the feedponint impedance at the load will you find EMax and EMin virtually anywhere along the length of the line.
similarly, if a transmitter source is connected to the input of an amplifier you have the same scenario. only when both the source output and the amplifier input present R=50 / X=0 will the length of the 50 ohm line in between them become unimportant as the ratio of EMax to EMin is the same all along the line.
After a bout with Doc and Beetle you could sometimes be worse off then when you first got started 
