kavkem numerals are based on a symmetric pattern cut into pieces, then assigned in order. To reconstruct the original pattern, 'connect' the bottom end of the arash numeral glyph to the left end of arap; the right end of arap 'connects' to arar; the bottom end of the araz numeral glyph 'connects' to the top of arab, and so forth.

The symmetric nature of the underlying pattern is why the first numeral, araz appears to have so much in common with the last numeral, araks. As long as you can commit the shape to memory, you should in theory be able to deduce the glyph for a numeral you want to express.

atel 0
araz 1		arash 2		aras 3		arar 4		aral 5
arah 6		arab 7		arap 8		aran 9		aram 10
arav 11		araph 12		arath 13		arat 14		arad 15
araj 16		arag 17		araq 18		arak 19		araks 20

The kavkem alphabet is larger than the Kendane͡ivash alphabet (which consists of the elements a, b, d, e, g, h, i, j, k, l, m, n, o, p, q, r, s, s̈, t, u, v, y, z), since possible digraphs have their own symbol, inspired by the numeral system where digraphs find direct use and get their own glyph as well.

It does, however, keep the Kendane͡ivash convention of keeping two separate alphabets - one for consonants, one for vowels. In recitations, they are often merged, with the shorter vowelphabet observing repetitions (zo, shu, sa, ri͡u, … ve, pho, thu, … s̈e, tso).

The consonant alphabet is derived from the numeral alphabet, with an added flourish, both to differentiate it from the numeral alphabet (to be able to use both numerals and letters in the same text without causing confusion) and as to be able to more easily write it.

z		sh		s		r		l
h		b		p		n		m
v		ph f		th		t		d
j		nj		g		q		k
ks		s̈		ts

The vowel alphabet is based on different shapes altogether.

o		u		a		i͡u
e͡i		i		i͡y		y
a͡i		u͡i		e