I would also add that measuring the spectrum of the signal as measured by a microphone is a bad idea to try to characterize its frequency response. I am assuming that that's what you did above.
ALERT: this is a boring and long summary of mic measurement thingies.
First of all, microphones are transducers. This means that they translates physical quantities between different domains. The microphone job is to translate pressure waves in air at one point in space into electric waves (signals). An ideal microphone would do this without boosting or attenuating any frequency: it would have a flat
response, at least in the audible range. Then, sensitivity
is a number that translates the input pressure amplitude into output voltage amplitude. Also speakers have sensitivity, but it goes the other way: from electric domain to pressure domain.
Now, microphones not only have one frequency response. They have infinite (!). I know it sounds weird, but every transducer have a certain directivity which depends on its construction. Without getting into details, at low frequency every transducer is pretty omnidirectional, so the response will be the same independently of ideal source position. As the frequency rises the traducers become more and more "directional", but side lobes develop. This picture
is about pistonic speakers, but it gives you an idea about what I am talking.
So, think about it: even just changing the test signal source position will alter the output spectrum you measure. Which is why microphone frequency responses are assessed only for normal incidence (source in front of the mic) and then the directivity is assessed by other methods.
Now, you are probably inside a room. Which has reflections. It does not matter whether the reverberation time is short: the shape of the room and acoustic impedance of the walls characterize completely the sound transmission law between 2 points within the room. Let's pick 2 points S and M anywhere in the room. Let's put a speaker in S and a microphone in M. Both speaker and microphone have a frequency response and directivity. This means that the signal at the microphone output will be the result of: response governing transduction from electric domain to acoustic domain in the speaker, response governing sound transmission from speaker to microphone position, response governing transduction to acoustic domain to electric domain. All of this passing through the combined effect of the speaker and microphone directivities, which might or might not attenuate or even completely remove room reflections coming from certain directions just depending on the microphone shape.
That is why measuring and calibrating microphones is very very hard, and done in special rooms (anechoic rooms). This goes as showing that your test is actually struggling in providing any actual properties of the mic under test. None of the curves are very realistic. Low end roll off is often found around 10 Hz, high end doesn't even roll off. It is made to roll off in electronics actually, as the high end of the transducer itself might have modes due to mic construction and membrane shape and material. These tend to be better to manage for condenser mics.
The only measure that it is reasonably easy to do at home (or normal rooms) is microphone response comparison. Pick two microphones you want to compare, and place them head to head, as close as possible but making sure they don't touch. Then, turn on a broad band noise source in the same room (white noise or pink noise are good). Then, collect the signal output from the microphones. Let's call the outputs x1 and x2. Go frequency domain with a Fourier transform, obtaining the frequency domain signals X1 X2. Then, divide to have the result: R = X2 / X1. The curve you obtain, R, is the difference in response between microphone 2 and microphone 1, which is usually called reference. Usually, what it is done is to pick a reference which is dead flat in frequency, so that the measurement above is very similar as measuring the actual response of the microphone under study.
Now, there is one thing that was not mentioned: linearity. A system is linear if its output is the result of only changing phase and amplitude of its input, even if differently depending on frequency. Nonlinear systems will also change the shape of the input signal. Condenser microphones are usually more linear, because the modes of their membranes can be made to happen easily outside the audible range (due to their usual geometry and mechanical properties). This means less distortion.