I have a problem understanding the following code:

PROGMEM  prog_uint16_t x={1232,3232,43343};
rawlen = pgm_read_word_near(x);
memcpy_P(uSendBuff, pfSendBuff+1, rawlen * sizeof(uint16_t));

Why do we need this? -

rawlen * sizeof(uint16_t));

Why not this? -

memcpy_P(uSendBuff, pfSendBuff+1, sizeof(rawlen));