Instead of trying to precisely measure the water isotopes in house standards against the VSMOW-SLAP scale, you can use a set of 3 appropriate house standards to mix your own VSMOW or SLAP standards and make sure, you get precisely the same readings as you get in the current international reference standards (VSMOW2, SLAP2), or mixtures of the two, no matter how good your calibration is.

Since the conversion from concentrations to the delta notation is linear, this is as easy as mixing waters of different concentrations. The only complication is that there is no water with zero concentrations of ¹⁸O or ²H. Even the strongly depleted SLAP standard with  δ¹⁸O = - 55.5  ‰ is equivalent to an ¹⁸O concentration of 1893.91 ppm and its δ ²H = - 427.5 ‰  is equivalent to a ²H concentration of 89.173 ppm.

Given you have standards A, B and C with the water isotopes  δ¹⁸O_A, δ²H_A,  δ¹⁸O_B, δ²H_B, δ¹⁸O_C, δ²H_C,
you may mix it in the fractions a, b and c to get a desired composition  δ¹⁸O_aim, δ²H_aim. using these equations:

....... conditions :

 (1)    a x δ¹⁸O_A + b x δ¹⁸O_B + c x δ¹⁸O_C =  δ¹⁸O_aim

 (2)    a x δ ²H_A + b x δ ²H_B + c x δ ²H_C =  δ ²H_aim

 (3)    a + b + c = 1

.......  auxiliary equations :

 (4 )    j =  δ ²H_A - δ ²H_B

 (5)    m = ( δ¹⁸O_aim - δ¹⁸O_A ) / ( δ¹⁸O_B - δ¹⁸O_A )

 (6)    n = ( δ¹⁸O_C - δ¹⁸O_A ) / (δ ¹⁸O_B - δ¹⁸O_A )

.........  result :

 (6)    c = ( δ ²H_aim - δ ²H_A + n x j ) / (δ ²H_C - δ ²H_A + n x j )

 (7)    b = m - n x c

 (8)    a = 1 - b - c

If you use the right selection of house standards, there is a possible solution, that gives you positive fractions for all three house standards. Obviously one has to be higher and one lower than the standard you are aiming to mix (otherwise you get negative fractions, which are impossible). And the third one is used to adjust the deuterium excess  (d = δ²H - 8 x δ¹⁸O) - meaning you have to choose an appropriate one, so it is actually elevating or decreasing your deuterium excess.

The attractive thing about this scheme is, that it is not necessary to have a precisely calibrated machine, you just need to know the exact reading for your house standards on your machine with your current calibration (or just uncalibrated machine readings), and the exact reading of the standard you are aiming to "copy".

Example: 

You have
solution A: δ¹⁸O_A,= + 2.1 ‰,  δ²H_A = + 13 ‰,       ( d_A = - 3.8 ‰ )
solution B: δ¹⁸O_B = - 6.7 ‰,   δ²H_B = - 50 ‰,       ( d_B = + 3.6 ‰ )
solution C: δ¹⁸O_C = - 13.0 ‰  δ²H_C = - 71 ‰,       ( d_C = + 33 ‰ )

you want to create VSMOW which shows in your machine with the settings you used for the other three solutions:

δ¹⁸O_aim = + 0.2 ‰  δ²H_aim = +1.4 ‰

you need a mix of 84.36 % solution 1,  7.34 % solution 2  and 8.31 % solution 3 !

δ¹⁸O_aim = 0.8436 x 2.1‰  + 0.0734 x (-6.7 ‰ ) + 0.0831 x (-13 ‰ ) = + 0.2 ‰
  δ²H_aim =  0.8436 x 13 ‰  + 0.0734 x (-50 ‰ )  + 0.0831 x (-71 ‰ ) = + 1.4 ‰

The other advantage is, that you are not relying on house standards, that might carry an error depending on the quality and linearity of the machine calibration. You simply try to get as close to the reference material as possible, regardless of its calibration. We had problems trying to use secondary standards that other labs had measured for us. We only found out our machine is actually very linear, when we prepared mixtures of the original standards VSMOW2 and SLAP2, which then have precisely known expected concentrations (at least to the precision you can trust your pipetting skills).

original time stamp 15.Feb 2018, changes 20Jan19